blob: 9c2f44fc98201d304ea54c386fa4df2d3f43fcf2 [file] [log] [blame]
C-Kermit Program Logic Manual
Frank da Cruz
[1]The Kermit Project
[2]Columbia University
As of: C-Kermit 8.0.211, 10 April 2004
This page last updated: Sat Apr 10 16:45:30 2004 (New York USA Time)
IF YOU ARE READING A PLAIN-TEXT version of this document, note that
this file is a plain-text dump of a Web page. You can visit the
original (and possibly more up-to-date) Web page here:
[ [4]C-Kermit Home ] [ [5]Kermit Home ]
2. [7]FILES
4.A. [10]Group A: Library Routines
4.B. [11]Group B: Kermit File Transfer
4.C. [12]Group C: Character-Set Conversion
4.D. [13]Group D: User Interface
4.E. [14]Group E: Platform-Dependent I/O
4.F. [15]Group F: Network Support
4.G. [16]Group G: Formatted Screen Support
4.H. [17]Group H: Pseudoterminal Support
4.I. [18]Group I: Security
The Kermit Protocol is specified in the book Kermit, A File Transfer
Protocol by Frank da Cruz, Digital Press / Butterworth Heinemann,
Newton, MA, USA (1987), 379 pages, ISBN 0-932376-88-6. It is assumed
the reader is familiar with the Kermit protocol specification.
This file describes the relationship among the modules and functions
of C-Kermit 5A and later, and other programming considerations.
C-Kermit is designed to be portable to any kind of computer that has a
C compiler. The source code is broken into many files that are grouped
according to their function, as shown in the [20]Contents.
C-Kermit has seen constant development since 1985. Throughout its
history, there has been a neverending tug-of-war among:
a. Functionality: adding new features, fixing bugs, improving
b. Adding support for new platforms.
c. "Buzzword 1.0 compliance".
The latter category is the most frustrating, since it generally
involves massive changes just to keep the software doing what it did
before in some new setting: e.g. the K&R-to-ANSIC conversion (which
had to be done, of course, without breaking K&R); Y2K (not a big deal
in our case); the many and varied UNIX and other API "standards";
[ [21]Contents ] [ [22]C-Kermit ] [ [23]Kermit Home ]
C-Kermit source files begin with the two letters "ck", for example
ckutio.c. Filenames are kept short (6.3) for maximum portability and
(obviously I hope) do not contain spaces or more than one period. The
third character in the name denotes something about the function group
and the expected level of portability:
a General descriptive material and documentation (text)
b BOO file encoders and decoders (obsolete)
c All platforms with C compilers (*)
d Data General AOS/VS
e Reserved for "ckermit" files, like ckermit.ini, ckermit2.txt
f (reserved)
g (reserved)
h (reserved)
i Commodore Amiga (Intuition)
j (unused)
k (unused)
l Stratus VOS
m Macintosh with Mac OS 1-9
n Microsoft Windows NT/2000/XP
o OS/2 and/or Microsoft Windows 9x/ME/NT/2000/XP
p Plan 9 from Bell Labs
q (reserved)
r DEC PDP-11 with RSTS/E (never used, open for reassigment)
s Atari ST GEMDOS (last supported in version 5A(189))
t DEC PDP-11 with RT-11 (never used, open for reassigment)
u Unix-based operating systems (*)
v VMS and OpenVMS
w Wart (Lex-like preprocessor, platform independent)
x (reserved)
y (reserved)
z (reserved)
0-3 (reserved)
4 IBM AS/400
5-8 (reserved)
9 Microware OS-9
_ Encryption modules
(*) In fact there is little distinction between the ckc*.* and cku*.*
categories. It would make more sense for all cku*.* modules to be
ckc*.* ones, except ckufio.c, ckutio.c, ckucon.c, ckucns.c, and
ckupty.c, which truly are specific to Unix. The rest (ckuus*.c,
ckucmd.c, etc) are quite portable.
One hint before proceeding: functions are scattered all over the
ckc*.c and cku*.c modules, where function size has begun to take
precedence over the desirability of grouping related functions
together, the aim being to keep any particular module from growing
disproportionately large. The easiest way (in UNIX) to find out in
what source file a given function is defined is like this (where the
desired function is foo()...):
grep ^foo\( ck*.c
This works because the coding convention has been to make function
names always start on the left margin with their contents indented,
for example:
static char *
foo(x,y) int x, y; {
Also note the style for bracket placement. This allows
bracket-matching text editors (such as EMACS) to help you make sure
you know which opening bracket a closing bracket matches, particularly
when the opening bracket is above the visible screen, and it also
makes it easy to find the end of a function (search for '}' on the
left margin).
Of course EMACS tags work nicely with this format too:
$ cd kermit-source-directory
$ etags ck[cu]*.c
$ emacs
Esc-X Visit-Tags-Table<CR><CR>
(but remember that the source file for ckcpro.c is [24]ckcpro.w!)
* Tabs should be set every 8 spaces, as on a VT100.
* All lines must no more than 79 characters wide after tab
* Note the distinction between physical tabs (ASCII 9) and the
indentation conventions, which are: 4 for block contents, 2 for
most other stuff (obviously this is not a portability issue, just
[ [25]Contents ] [ [26]C-Kermit ] [ [27]Kermit Home ]
C-Kermit was designed in 1985 as a platform-independent replacement
for the earlier Unix Kermit. c-Kermit's design was expected to promote
portability, and judging from the number of platforms to which it has
been adapted since then, the model is effective, if not ideal
(obviously if we had it all to do over, we'd change a few things). To
answer the oft-repeated question: "Why are there so many #ifdefs?",
it's because:
* Many of them are related to feature selection and program size,
and so need to be there anyway.
* Those that treat compiler, library, platform, header-file, and
similar differences have built up over time as hundreds of people
all over the world adapted C-Kermit to their particular
environments and sent back their changes. There might be more
politically-correct ways to achieve portability, but this one is
natural and proven. The basic idea is to introduce changes that
can be selected by defining a symbol, which, if not defined,
leaves the program exactly as it was before the changes.
* Although it might be possible to "clean up" the "#ifdef mess",
nobody has access to all the hundreds of platforms served by the
#ifdefs to check the results.
And to answer the second-most-oft-repeated question: "Why don't you
just use GNU autoconfig / automake / autowhatever instead of
hard-coding all those #ifdefs?" Answers:
* The GNU tools are not available on all the platforms where
C-Kermit must be built and I wouldn't necessarily trust them if
they were.
* Each platform is a moving target, so the tools themselves would
need to updated before Kermit could be updated.
* It would only add another layer of complexity to an already
complex process.
* Conversion at this point would not be practical unless there was a
way to test the results on all the hundreds of platforms where
C-Kermit is supposed to build.
When writing code for the system-indendent C-Kermit modules, please
stick to the following coding conventions to ensure portability to the
widest possible variety of C preprocessors, compilers, and linkers, as
well as certain network and/or email transports. The same holds true
for many of the "system dependent" modules too; particularly the Unix
ones, since they must be buildable by a wide variety of compilers and
linkers, new and old.
This list does not purport to be comprehensive, and although some
items on it might seem far-fetched, they would not be listed unless I
had encountered them somewhere, some time. I wish I had kept better
records so I could cite specific platforms and compilers.
* Try to keep variable and function names unique within 6
characters, especially if they are used across modules, since 6 is
the maximum for some old linkers (actually, this goes back to
TOPS-10 and -20 and other old DEC OS's where C-Kermit never ran
anyway; a more realistic maximum is probably somewhere between 8
and 16). We know for certain that VAX C has a 31-character max
because it complains -- others might not complain, but just
silently truncate, thus folding two or more routines/variables
into one.
* Keep preprocessor symbols unique within 8 characters; that's the
max for some preprocessors (sorry, I can't give a specific
example, but in 1988 or thereabouts, I had to change character-set
symbols like TC_LATIN1 and TC_LATIN2 to TC_1LATIN and TC_2LATIN
because the digits were being truncated and ignored on a platform
where I actually had to build C-Kermit 5A; unfortunately I didn't
note which platform -- maybe some early Ultrix version?)
* Don't create preprocessor symbols, or variable or function names,
that start with underscore (_). These are usually reserved for
internal use by the compiler and header files.
* Don't put #include directives inside functions or { blocks }.
* Don't use the #if or #elif preprocessor constructions, only use
#ifdef, #ifndef, #define, #undef, and #endif.
* Put tokens after #endif in comment brackets, e.g.
#endif /* FOO */.
* Don't indent preprocessor statements - # must always be first char
on line.
* Don't put whitespace after # in preprocessor statements.
* Don't use #pragma, even within #ifdefs -- it makes some
preprocessors give up.
* Same goes for #module, #if, etc - #ifdefs do NOT protect them.
* Don't use logical operators in preprocessor constructions.
* Avoid #ifdefs inside argument list to function calls (I can't
remember why this one is here, but probably needn't be; we do this
all the time).
* Always cast strlen() in expressions to int:
if ((int)strlen(foo) < x)...
* Any variable whose value might exceed 16383 should be declared as
long, or if that is not possible, then as unsigned.
* Avoid typedefs; they might be portable but they are very confusing
and there's no way to test for their presence or absence at
compile time. Use preprocessor symbols instead if possible; at
least you can test their definitions.
* Unsigned long is not portable; use a preprocessor symbol (Kermit
uses ULONG for this).
* Long long is not portable. If you really need it, be creative.
* Similarly 1234LL is not portable, nor almost any other constant
modifier other than L.
* Unsigned char is not portable, use CHAR (a preprocessor symbol
defined in the Kermit header files) and always take precautions
against character signage (more about this [28]below).
* Don't use initializers with automatic arrays or structs: it's not
* Don't use big automatic arrays or structs in functions that might
be called recursively; some platforms have fixed-size stacks (e.g.
Windows 9x: 256K) and recursive functions crash with stack
overflow. Even when there is not a compiler limitation, this
causes memory to be consumed without bound, and can end up filling
swap space.
* Don't assume that struct assignment performs a copy, or that it
even exists.
* Don't use sizeof to get the size of an array; someone might come
along later and and change it from static to malloc'd. Always use
a symbol to refer to the array's size.
* Don't put prototypes for static functions into header files that
are used by modules that don't contain that function; the link
step can fail with unresolved references (e.g. on AOS/VS).
* Avoid the construction *++p (the order of evaluation varies; it
shouldn't but at least one compiler had a bug that made me include
this item).
* Don't use triple assignments, like a = b = c = 0; (or quadruple,
etc). Some compilers generate bad code for these, or crash, etc
(some version of DEC C as I recall).
* Some compilers don't allow structure members to have the same
names as other identifiers. Try to give structure members unique
* Don't assume anything about order of evaluation in boolean
expressions, or that they will stop early if a required condition
is not true, e.g.:
if (i > 0 && p[i-1] == blah)
can still dump core if i == 0 (hopefully this is not true of any
modern compiler, but I would not have said this if it did not
actually happen somewhere).
* Don't have a switch() statement with no cases (e.g. because of
#ifdefs); this is a fatal error in some compilers.
* Don't put lots of code in a switch case; move it out to a separate
function; some compilers run out of memory when presented with a
huge switch() statement -- it's not the number of cases that
matters; it's the overall amount of code.
* Some compilers might also limit the number of switch() cases, e.g.
to 254.
* Don't put anything between "switch() {" and "case:" -- switch
blocks are not like other blocks.
* Don't jump into or out of switches.
* Don't make character-string constants longer than about 250 bytes.
Longer strings should be broken up into arrays of strings.
* Don't write into character-string constants (obviously). Even when
you know you are not writing past the end; the compiler or linker
might have put them into read-only and/or shared memory, and/or
coalesced multiple equal constants so if you change one you change
them all.
* Don't depend on '\r' being carriage return.
* Don't depend on '\n' being linefeed or for that matter any SINGLE
* Don't depend on '\r' and '\n' being different (e.g. as separate
switch() cases).
* In other words, don't use \n or \r to stand for specific
characters; use \012 and \015 instead.
* Don't code for "buzzword 1.0 compliance", unless "buzzword" is K&R
and "1.0" is the first edition.
* Don't use or depend on anything_t (size_t, pid_t, etc), except
time_t, without #ifdef protection (time_t is the only one I've
found that is accepted everywhere). This is a tough one because
the same function might require (say) a size_t arg on one
platform, whereas size_t is unheard of on another; or worse, it
might require a totally different data type, like int or long or
some other typedef'd thing. It has often proved necessary to
define a symbol to stand for the type of a particular argument to
a particular library or system function to get around this
* Don't use or depend on internationalization ("i18n") features,
wchar_t, locales, etc, in portable code; they are not portable.
Anyway, locales are not the right model for Kermit's
multi-character-set support. Kermit does all character-set
conversion itself and does not use any external libraries or
* In particular, don't use any library functions that deal with wide
characters or Unicode in any form. These are not only nonportable,
but a constantly shifting target (e.g. the ones in glibc).
* Don't make any assumption about signal handler type. It can be
void, int, long, or anything else. Always declare signal handlers
as SIGTYP (see definition in ckcdeb.h and augment it if necessary)
and always use SIGRETURN at exit points from signal handlers.
* Signals should always be re-armed to be used again (this barely
scratches the surface -- the differences between BSD/V7 and System
V and POSIX signal handling are numerous, and some platforms do
not even support signals, alarms, or longjmps correctly or at all
-- avoid all of this if you can).
* On the other hand, don't assume that signals are disarmed after
being raised. In some platforms you have to re-arm them, in others
they stay armed.
* Don't call malloc() and friends from a signal handler; don't do
anything but setting integer global variables in a signal handler.
* malloc() does not initialize allocated memory -- it never said it
did. Don't expect it to be all 0's.
* Did You Know: malloc() can succeed and the program can still dump
core later when it attempts to use the malloc'd memory? (This
happens when allocation is deferred until use and swap space is
* memset(), memmove(), and memcpy() are not portable, don't use them
without protecting them in ifdefs (we have USE_MEMCPY for this).
bzero()/bcopy() too, except we're guaranteed to have
bzero()/bcopy() when using the sockets library (not really). See
examples in the source.
* Don't assume that strncpy() stops on the first null byte -- most
versions always copy the number of bytes given in arg 3, padding
out with 0's and overwriting whatever was there before. Use
C-Kermit ckstrncpy() if you want predictable non-padding behavior,
guaranteed NUL-termination, and a useful return code.
* DID YOU KNOW.. that some versions of inet_blah() routines return
IP addresses in network byte order, while others return them local
machine byte order? So passing them to ntohs() or whatever is not
always the right thing to do.
* Don't use ANSI-format function declarations without #ifdef
CK_ANSIC, and always provide an #else for the non-ANSI case.
* Use the Kermit _PROTOTYP() macro for declaring function
prototypes; it works in both the ANSI and non-ANSI cases.
* Don't depend on any other ANSI preprocessor features like
"pasting" -- they are often missing or nonoperational.
* Don't assume any C++ syntax or semantics.
* Don't use // as a comment introducer. C is not C++.
* Don't declare a string as "char foo[]" in one module and "extern
char * foo" in another, or vice-versa: this causes core dumps.
* With compiler makers falling all over themselves trying to outdo
each other in ANSI strictness, it has become increasingly
necessary to cast EVERYTHING. This is increasingly true for char
vs unsigned char. We need to use unsigned chars if we want to deal
with 8-bit character sets, but most character- and string-oriented
APIs want (signed) char arguments, so explicit casts are
necessary. It would be nice if every compiler had a
-funsigned-char option (as gcc does), but they don't.
* a[x], where x is an unsigned char, can produce a wild memory
reference if x, when promoted to an int, becomes negative. Cast it
to (unsigned), even though it ALREADY IS unsigned.
* Be careful how you declare functions that have char or long
arguments; for ANSI compilers you MUST use ANSI declarations to
avoid promotion problems, but you can't use ANSI declarations with
non-ANSI compilers. Thus declarations of such functions must be
hideously entwined in #ifdefs. Example: latter:
int /* Put character in server command buffer */
#ifdef CK_ANSIC
putsrv(char c)
putsrv(c) char c;
#endif /* CK_ANSIC */
/* putsrv */ {
*srvptr++ = c;
*srvptr = '\0'; /* Make sure buffer is null-terminated */
* Be careful how you return characters from functions that return
int values -- "getc-like functions" -- in the ANSI world. Unless
you explicitly cast the return value to (unsigned), it is likely
to be "promoted" to an int and have its sign extended.
* At least one compiler (the one on DEC OSF/1 1.3) treats "/*" and
"*/" within string constants as comment begin and end. No amount
of #ifdefs will get around this one. You simply can't put these
sequences in a string constant, e.g. "/usr/local/doc/*.*".
* Avoid putting multiple macro references on a single line, e.g.:
putchar(BS); putchar(SP); putchar(BS)
This overflows the CPP output buffer of more than a few C
preprocessors (this happened, for example, with SunOS 4.1 cc, which
evidently has a 1K macro expansion buffer).
C-Kermit needs constant adjustment to new OS and compiler releases.
Every new OS release shuffles header files or their contents, or
prototypes, or data types, or levels of ANSI strictness, etc. Every
time you make an adjustment to remove a new compilation error, BE VERY
CAREFUL to #ifdef it on a symbol unique to the new configuration so
that the previous configuration (and all other configurations on all
other platforms) remain as before.
Assume nothing. Don't assume header files are where they are supposed
to be, that they contain what you think they contain, that they define
specific symbols to have certain values -- or define them at all!
Don't assume system header files protect themselves against multiple
inclusion. Don't assume that particular system or library calls are
available, or that the arguments are what you think they are -- order,
data type, passed by reference vs value, etc. Be conservative when
attempting to write portable code. Avoid all advanced features.
If you see something that does not make sense, don't assume it's a
mistake -- it might be there for a reason, and changing it or removing
is likely to cause compilation, linking, or runtime failures sometime,
somewhere. Some huge percentage of the code, especially in the
platform-dependent modules, is workarounds for compiler, linker, or
API bugs.
But finally... feel free to violate any or all of these rules in
platform-specific modules for environments in which the rules are
certain not to apply. For example, in VMS-specific code, it is OK to
use #if, because VAX C, DEC C, and VMS GCC all support it.
[ [29]Contents ] [ [30]C-Kermit ] [ [31]Kermit Home ]
3.1. Memory Leaks
The C language and standard C library are notoriously inadequate and
unsafe. Strings are arrays of characters, usually referenced through
pointers. There is no native string datatype. Buffers are fixed size,
and C provides no runtime bounds checking, thus allowing overwriting
of other data or even program code. With the popularization of the
Internet, the "buffer exploit" has become a preferred method for
hackers to hijack privileged programs; long data strings are fed to a
program in hopes that it uses unsafe C library calls such as strcpy()
or sprintf() to copy strings into automatic arrays, thus overwriting
the call stack, and therefore the routine's return address. When such
a hole is discovered, a "string" can be constructed that contains
machine code to hijack the program's privileges and penetrate the
This problem is partially addressed by the strn...() routines, which
should always be used in preference to their str...() equivalents
(except when the copy operation has already been prechecked, or there
is a good reason for not using them, e.g. the sometimes undesirable
side effect of strncpy() zeroing the remainder of the buffer). The
most gaping whole, however, is sprintf(), which performs no length
checking on its destination buffer, and is not easy to replace.
Although snprintf() routines are starting to appear, they are not yet
widespread, and certainly not universal, nor are they especially
portable, or even full-featured.
For these reasons, we have started to build up our own little library
of C Library replacements, ckclib.[ch]. These are safe and highly
portable primitives for memory management and string manipulation,
such as:
Like strncpy but returns a useful value, doesn't zero buffer.
Opposite of atoi()
Opposite of atol()
Returns character as string
Used with ck?to?() as a safe sprintf() replacement for up to 4
Like ckmakmsg() but accepts up to 12 items
More about library functions in [32]Section 4.A.
[ [33]Contents ] [ [34]C-Kermit ] [ [35]Kermit Home ]
3.2. The "char" vs "unsigned char" Dilemma
This is one of the most aggravating and vexing characteristics of the
C language. By design, chars (and char *'s) are SIGNED. But in the
modern era, however, we need to process characters that can have (or
include) 8-bit values, as in the ISO Latin-1, IBM CP 850, or UTF-8
character sets, so this data must be treated as unsigned. But some C
compilers (such as those based on the Bell UNIX V7 compiler) do not
support "unsigned char" as a data type. Therefore we have the macro or
typedef CHAR, which we use when we need chars to be unsigned, but
which, unfortunately, resolves itself to "char" on those compilers
that don't support "unsigned char". AND SO... We have to do a lot of
fiddling at runtime to avoid sign extension and so forth.
Some modern compilers (e.g. IBM, DEC, Microsoft) have options that say
"make all chars be unsigned" (e.g. GCC "-funsigned-char") and we use
them when they are available. Other compilers don't have this option,
and at the same time, are becoming increasingly strict about type
mismatches, and spew out torrents of warnings when we use a CHAR where
a char is expected, or vice versa. We fix these one by one using
casts, and the code becomes increasingly ugly. But there remains a
serious problem, namely that certain library and kernel functions have
arguments that are declared as signed chars (or pointers to them),
whereas our character data is unsigned. Fine, we can can use casts
here too -- but who knows what happens inside these routines.
[ [36]Contents ] [ [37]C-Kermit ] [ [38]Kermit Home ]
When C-Kermit is on the far end of a connection, it is said to be in
remote mode. When C-Kermit has made a connection to another computer,
it is in local mode. (If C-Kermit is "in the middle" of a multihop
connection, it is still in local mode.)
On another axis, C-Kermit can be in any of several major states:
Command State
Reading and writing from the job's controlling terminal or
"console". In this mode, all i/o is handled by the Group E
conxxx() (console i/o) routines.
Protocol State
Reading and writing from the communicatons device. In this
mode, all i/o is handled by the Group E ttxxx() (terminal i/o)
Terminal State
Reading from the keyboard with conxxx() routines and writing to
the communications device with ttxxx() routines AND vice-versa.
When in local mode, the console and communications device are
distinct. During file transfer, Kermit may put up a file-transfer
display on the console and sample the console for interruption
When in remote mode, the console and communications device are the
same, and therefore there can be no file-transfer display on the
console or interruptions from it (except for "in-band" interruptions
such as ^C^C^C).
[ [39]Contents ] [ [40]C-Kermit ] [ [41]Kermit Home ]
4.A. Group A: Library Functions
Library functions, strictly portable, can be used by all modules on
all platforms: [42]ckclib.h, [43]ckclib.c.
(To be filled in... For now, see [44]Section 3.1 and the comments in
[ [45]Contents ] [ [46]C-Kermit ] [ [47]Kermit Home ]
4.B. Group B: Kermit File Transfer
The Kermit protocol kernel. These files, whose names start with "ckc
are supposed to be totally portable C, and are expected to compile
correctly on any platform with any C compiler. "Portable" does not
mean the same as as "ANSI" -- these modules must compile on 10- and
20-year old computers, with C preprocessors, compilers, and/or linkers
that have all sorts of restrictions. The Group B modules do not
include any header files other than those that come with Kermit
itself. They do not contain any library calls except from the standard
C library (e.g. printf()). They most certainly do not contain any
system calls. Files:
For use by C compilers that don't allow -D on the command line.
ASCII character symbol definitions.
System-independent signal-handling definitions and prototypes.
Originally, debugging definitions. Now this file also contains
all definitions and prototypes that are shared by all modules
in all groups.
Kermit protocol symbol definitions.
Character-set-related symbol definitions (see next section).
The main program. This module contains the declarations of all
the protocol-related global variables that are shared among the
other modules.
The protocol module itself, written in "wart", a lex-like
preprocessor that is distributed with Kermit under the name
[56]ckcfns.c, [57]ckcfn2.c, [58]ckcfn3.c
The protocol support functions used by the protocol module.
[59]Group B modules may call upon functions from [60]Group E, but not
from [61]Group D modules (with the single exception that the main
program invokes the user interface, which is in Group D). (This last
assertion is really only a conjecture.)
[ [62]Contents ] [ [63]C-Kermit ] [ [64]Kermit Home ]
4.C. Group C: Character-Set Conversion
Character set translation tables and functions. Used by the [65]Group
B, protocol modules, but may be specific to different computers. (So
far, all character character sets supported by C-Kermit are supported
in [66]ckuxla.c and [67]ckuxla.h, including Macintosh and IBM
character sets). These modules should be completely portable, and not
rely on any kind of system or library services.
Character-set definitions usable by all versions of C-Kermit.
Character-set definitions for computer "?", e.g. [69]ckuxla.h
for UNIX, [70]ckmxla.h for Macintosh.
Character-set translation tables and functions for computer
"?", For example, CKUXLA.C for UNIX, CKMXLA.C for Macintosh. So
far, these are the only two such modules. The UNIX module is
used for all versions of C-Kermit except the Macintosh version.
Unicode definitions
Unicode module
Here's how to add a new file character set in the original
(non-Unicode modules). Assuming it is based on the Roman (Latin)
alphabet. Let's call it "Barbarian". First, in ck?xla.h, add a
definition for FC_BARBA (8 chars maximum length) and increase
MAXFCSETS by 1. Then, in ck?xla.c:
* Add a barbarian entry into the fcsinfo array.
* Add a "barbarian" entry to file character set keyword table,
* Add a "barbarian" entry to terminal character set keyword table,
* Add a translation table from Latin-1 to barbarian: yl1ba[].
* Add a translation table from barbarian to Latin-1: ybal1[].
* Add a translation function from Barbarian to ASCII: xbaas().
* Add a translation function from Barbarian to Latin-1: xbal1().
* Add a translation function from Latin-1 to Barbarian: xl1ba().
* etc etc for each transfer character set...
* Add translation function pointers to the xls and xlr tables.
Other translations involving Barbarian (e.g. from Barbarian to
Latin-Cyrillic) are performed through these tables and functions. See
ckuxla.h and ckuxla.c for extensive examples.
To add a new Transfer Character Set, e.g. Latin Alphabet 9 (for the
Euro symbol), again in the "old" character-set modules:
In ckcxla.h:
+ Add a TC_xxxx definition and increase MAXTCSETS accordingly.
In ck?xla.h (since any transfer charset is also a file charset):
+ Add an FC_xxxx definition and increase MAXFCSETS accordingly.
In ck?xla.c:
+ Add a tcsinfo[] entry.
+ Make a tcstab[] keyword table entry.
+ Make an fcsinfo[] table entry.
+ Make an fcstab[] keyword table entry.
+ Make a tcstab[] keyword table entry.
+ If necessary, make a langinfo[] table entry.
+ Make entries in the function pointer arrays.
+ Provide any needed functions.
As of C-Kermit 7.0, character sets are also handled in parallel by the
new (and very large) Unicode module, ckcuni.[ch]. Eventually we should
phase out the old way, described just above, and operate entirely in
(and through) Unicode. The advantages are many. The disadvantages are
size and performance. To add a character to the Unicode modules:
In ckcuni.h:
+ (To be filled in...)
In ckcuni.c:
+ (To be filled in...)
[ [74]Contents ] [ [75]C-Kermit ] [ [76]Kermit Home ]
4.D. Group D: User Interface
This is the code that communicates with the user, gets her commands,
informs her of the results. It may be command-line oriented,
interactive prompting dialog, menus and arrow keys, windows and mice,
speech recognition, telepathy, etc. The one provided is command-and
prompt, with the ability to read commands from various sources: the
console keyboard, a file, or a macro definition. The user interface
has three major functions:
1. Sets the parameters for the file transfer and then starts it. This
is done by setting certain (many) global variables, such as the
protocol machine start state, the file specification, file type,
communication parameters, packet length, window size, character
set, etc.
2. Displays messages on the user's screen during the file transfer,
using the screen() function, which is called by the group-1
3. Executes any commands directly that do not require Kermit
protocol, such as the CONNECT command, local file management
commands, parameter-setting commands, FTP client commands, etc.
If you plan to imbed the [77]Group B, files into a program with a
different user interface, your interface must supply an appropriate
screen() function, plus a couple related ones like chkint() and
intmsg() for handling keyboard (or mouse, etc) interruptions during
file transfer. The best way to find out about this is to link all the
C-Kermit modules together except the ckuu*.o and ckucon.o modules, and
see which missing symbols turn up.
C-Kermit's character-oriented user interface (as opposed to the
Macintosh version's graphical user interface) consists of the
following modules. C-Kermit can be built with an interactive command
parser, a command-line-option-only parser, a graphical user interface,
or any combination, and it can even be built with no user interface at
all (in which case it runs as a remote-mode Kermit server).
The command parsing primitives used by the interactive command
parser to parse keywords, numbers, filenames, etc, and to give
help, complete fields, supply defaults, allow abbreviations and
editing, etc. This package is totally independent of Kermit,
but does depend on the [80]Group E functions.
Definitions of symbols used in Kermit's commands.
Kermit's interactive command parser, including the script
programming language: [82]ckuusr.c (includes top-level keyword
tables); [83]ckuus2.c (HELP command text); [84]ckuus3.c (most
of the SET command); [85]ckuus4.c (includes variables and
functions); ckuus[567].c (miscellaneous);
The command-line-option parser.
User interface functions common to both the interactive and
command-line parsers.
Version heralds for different implementations.
The (old, uucp-like) SCRIPT command
The DIAL command. Includes specific knowledge of many types of
Note that none of the above files is actually Unix-specific. Over time
they have proven to be portable among all platforms where C-Kermit is
built: Unix, VMS, AOS/VS, Amiga, OS-9, VOS, etc etc. Thus the third
letter should more properly be "c", but changing it would be too
ck?con.c, ckucns.c
The CONNECT command. Terminal connection, and in some cases
(Macintosh, Windows) also terminal emulation. NOTE: As of
C-Kermit 7.0, there are two different CONNECT modules for UNIX:
[91]ckucon.c -- the traditional, portable, fork()-based version
-- and [92]ckucns.c, a new version that uses select() rather
than forks so it can handle encryption. ckucns.c is the
preferred version for Unix; ckucon.c is not likely to keep pace
with it in terms of upgrades, etc. However, since select() is
not portable to every platform, ckucon.c will be kept
indefinitely for those platforms that can't use ckucns.c. NOTE:
SunLink X.25 support is available only in ckucon.c.
ck_*.*, ckuat*.*
Modules having to do with authentication and encryption. Since
the relaxation of USA export laws, they are included with the
general source-code distribution. Secure C-Kermit binaries can
be built using special targets in the standard makefile.
However, secure prebuilt binaries may not be distributed.
For other implementations, the files may, and probably do, have
different names. For example, the Macintosh graphical user interface
filenames start with "ckm". Kermit 95 uses the ckucmd and ckuus*
modules, but has its own CONNECT command modules. And so on.
Here is a brief description of C-Kermit's "user interface interface",
from ckuusr.c. It is nowhere near complete; in particular, hundreds of
global variables are shared among the many modules. These should, some
day, be collected into classes or structures that can be passed around
as needed; not only for purity's sake, but also to allow for multiple
simultaneous communication sessions and or user interfaces. Our list
of things to do is endless, and reorganizing the source is almost
always at the bottom.
The ckuus*.c modules (like many of the ckc*.c modules) depend on the
existence of C library features like fopen, fgets, feof, (f)printf,
argv/argc, etc. Other functions that are likely to vary among
operating systems -- like setting terminal modes or interrupts -- are
invoked via calls to functions that are defined in the [93]Group E
platform-dependent modules, ck?[ft]io.c. The command line parser
processes any arguments found on the command line, as passed to main()
via argv/argc. The interactive parser uses the facilities of the cmd
package (developed for this program, but, in theory, usable by any
program). Any command parser may be substituted for this one. The only
requirements for the Kermit command parser are these:
1. Set parameters via global variables like duplex, speed, ttname,
etc. See [94]ckcmai.c for the declarations and descriptions of
these variables.
2. If a command can be executed without the use of Kermit protocol,
then execute the command directly and set the sstate (start state)
variable to 0. Examples include SET commands, local directory
listings, the CONNECT command.
3. If a command requires the Kermit protocol, set the following
sstate string data
'x' (enter server mode) (none)
'r' (send a 'get' command) cmarg, cmarg2
'v' (enter receive mode) cmarg2
'g' (send a generic command) cmarg
's' (send files) nfils, cmarg & cmarg2 OR cmlist
'c' (send a remote host command) cmarg
cmlist is an array of pointers to strings.
cmarg, cmarg2 are pointers to strings.
nfils is an integer (hmmm, probably should be an unsigned long).
cmarg can be:
A filename string (possibly wild), or:
a pointer to a prefabricated generic command string, or:
a pointer to a host command string.
cmarg2 is:
The name to send a single file under, or:
the name under which to store an incoming file; must not
be wild.
If it's the name for receiving, a null value means to
store the file under the name it arrives with.
cmlist is:
A list of nonwild filenames, such as passed via argv.
nfils is an integer, interpreted as follows:
-1: filespec (possibly wild) in cmarg, must be expanded
0: send from stdin (standard input).
>0: number of files to send, from cmlist.
The screen() function is used to update the screen during file
transfer. The tlog() function writes to a transaction log (if TLOG is
defined). The debug() function writes to a debugging log (if DEBUG is
defined). The intmsg() and chkint() functions provide the user i/o for
interrupting file transfers.
[ [95]Contents ] [ [96]C-Kermit ] [ [97]Kermit Home ]
4.E. Group E: Platform-Dependent I/O
Platform-dependent function definitions. All the Kermit modules,
including the command package, call upon these functions, which are
designed to provide system-independent primitives for controlling and
manipulating devices and files. For Unix, these functions are defined
in the files [98]ckufio.c (files), [99]ckutio.c (communications), and
[100]ckusig.c (signal handling).
For VMS, the files are [101]ckvfio.c, ckvtio.c, and [102]ckusig.c (VMS
can use the same signal handling routines as Unix). It doesn't really
matter what the files are called, except for Kermit distribution
purposes (grouping related files together alphabetically), only that
each function is provided with the name indicated, observes the same
calling and return conventions, and has the same type.
The Group E modules contain both functions and global variables that
are accessed by modules in the other groups. These are now described.
(By the way, I got this list by linking all the C-Kermit modules
together except ckutio and ckufio. These are the symbols that ld
reported as undefined. But that was a long time ago, probably circa
Version 6.)
4.E.1. Global Variables
char *DELCMD;
Pointer to string containing command for deleting files.
Example: char *DELCMD = "rm -f "; (UNIX)
Example: char *DELCMD = "delete "; (VMS)
Note trailing space. Filename is concatenated to end of this
string. NOTE: DELCMD is used only in versions that do not
provide their own built-in DELETE command.
char *DIRCMD;
Pointer to string containing command for listing files when a
filespec is given.
Example: char *DIRCMD = "/bin/ls -l "; (UNIX)
Example: char *DIRCMD = "directory "; (VMS)
Note trailing space. Filename is concatenated to end of this
string. NOTE: DIRCMD is used only in versions that do not
provide their own built-in DIRECTORY command.
char *DIRCM2;
Pointer to string containing command for listing files when a
filespec is not given. (currently not used, handled in another
Example: char *DIRCMD2 = "/bin/ls -ld *";
NOTE: DIRCMD2 is used only in versions that do not provide
their own built-in DIRECTORY command.
char *PWDCMD;
Pointer to string containing command to display current
Example: char *PWDCMD = "pwd ";
NOTE: PWDCMD is used only in versions that do not provide their
own built-in PWD command.
char *SPACMD;
Pointer to command to display free disk space in current
Example: char *SPACMD = "df .";
NOTE: SPACMD is used only in versions that do not provide their
own built-in SPACE command.
char *SPACM2;
Pointer to command to display free disk space in another
Example: char *SPACM2 = "df ";
Note trailing space. Device or directory name is added to this
string. NOTE: SPACMD2 is used only in versions that do not
provide their own built-in SPACE command.
char *TYPCMD;
Pointer to command for displaying the contents of a file.
Example: char *TYPCMD = "cat ";
Note trailing space. Device or directory name is added to this
string. NOTE: TYPCMD is used only in versions that do not
provide their own built-in TYPE command.
char *WHOCMD;
Pointer to command for displaying logged-in users.
Example: char *WHOCMD = "who ";
Note trailing space. Specific user name may be added to this
int backgrd = 0;
Flag for whether program is running in foreground (0) or
background (nonzero). Background operation implies that screen
output should not be done and that all errors should be fatal.
int ckxech;
Flag for who is to echo console typein:
1: The program (system is not echoing).
0: The OS, front end, terminal, etc (not this program).
char *ckxsys;
Pointer to string that names the computer and operating system.
Example: char *ckxsys = " NeXT Mach 1.0";
Tells what computer system ckxv applies to. In UNIX Kermit,
this variable is also used to print the program herald, and in
the SHOW VERSION command.
char *ckxv;
Pointer to version/edit info of ck?tio.c module.
Example: char *ckxv = "UNIX Communications Support, 6.0.169, 6
Sep 96";
Used by SHOW VERSION command.
char *ckzsys;
Like ckxsys, but briefer.
Example: char *ckzsys = " 4.3 BSD";
Tells what platform ckzv applies to. Used by the SHOW VERSION
char *ckzv;
Pointer to version/edit info of ck?fio.c module.
Example: char *ckzv = "UNIX File support, 6.0.113, 6 Sep 96";
Used by SHOW VERSION command.
int dfflow;
Default flow control. 0 = none, 1 = Xon/Xoff, ... (see FLO_xxx
symbols in ckcdeb.h)
Set by Group E module. Used by [103]ckcmai.c to initialize flow
control variable.
int dfloc;
Default location. 0 = remote, 1 = local. Set by Group E module.
Used by ckcmai.c to initialize local variable. Used in various
places in the user interface.
int dfprty;
Default parity. 0 = none, 'e' = even, 'o' = odd, 'm' = mark,
's' = space. Set by Group E module. Used by ckcmai.c to
initialize parity variable.
char *dftty;
Default communication device. Set by Group E module. Used in
many places. This variable should be initialized the the symbol
CTTNAM, which is defined in ckcdeb.h, e.g. as "/dev/tty" for
UNIX, "TT:" for VMS, etc. Example: char *dftty = CTTNAM;
char *mtchs[];
Array of string pointers to filenames that matched the most
recent wildcard match, i.e. the most recent call to zxpand().
Used (at least) by command parsing package for partial filename
int tilde_expand;
Flag for whether to attempt to expand leading tildes in
directory names (used in UNIX only, and then only when the
symbol DTILDE is defined.
int ttnproto;
The protocol being used to communicate over a network device.
Values are defined in ckcnet.h. Example: NP_TELNET is network
protocol "telnet".
int maxnam;
The maximum length for a filename, exclusive of any device or
directory information, in the format of the host operating
int maxpath;
The maximum length for a fully specified filename, including
device designator, directory name, network node name, etc, in
the format of the host operating system, and including all
int ttyfd;
File descriptor of the communication device. -1 if there is no
open or usable connection, including when C-Kermit is in remote
mode. Since this is not implemented everywhere, references to
it are in #ifdef CK_TTYFD..#endif.
[ [104]Contents ] [ [105]C-Kermit ] [ [106]Kermit Home ]
4.E.2. Functions
These are divided into three categories: file-related functions (B.1),
communication functions (B.2), and miscellaneous functions (B.3).
4.E.2.1. File-Related Functions
In most implementations, these are collected together into a module
called ck?fio.c, where ? = "u" ([107]ckutio.c for Unix), "v"
([108]ckvtio.c for VMS), [109]etc. To be totally platform-independent,
C-Kermit maintains its own file numbers, and provides the functions
described in this section to deal with the files associated with them.
The file numbers are referred to symbolically, and are defined as
follows in ckcker.h:
#define ZCTERM 0 /* Console terminal */
#define ZSTDIO 1 /* Standard input/output */
#define ZIFILE 2 /* Current input file for SEND command */
#define ZOFILE 3 /* Current output file for RECEIVE command */
#define ZDFILE 4 /* Current debugging log file */
#define ZTFILE 5 /* Current transaction log file */
#define ZPFILE 6 /* Current packet log file */
#define ZSFILE 7 /* Current session log file */
#define ZSYSFN 8 /* Input from a system function (pipe) */
#define ZRFILE 9 /* Local file for READ command */ (NEW)
#define ZWFILE 10 /* Local file for WRITE command */ (NEW)
#define ZMFILE 11 /* Auxilliary file for internal use */ (NEW)
#define ZNFILS 12 /* How many defined file numbers */
In the descriptions below, fn refers to a filename, and n refers to
one of these file numbers. Functions are of type int unless otherwise
noted, and are listed mostly alphabetically.
chkfn(n) int n;
Checks the file number n. Returns:
-1: File number n is out of range
0: n is in range, but file is not open
1: n in range and file is open
iswild(filspec) char *filespec;
Checks if the file specification is "wild", i.e. contains
metacharacters or other notations intended to match multiple
filenames. Returns:
0: not wild
1: wild.
isdir(string) char *string;
Checks if the string is the name of an existing directory. The
idea is to check whether the string can be "cd'd" to, so in
some cases (e.g. DOS) it might also indicate any file
structured device, such as a disk drive (like A:). Other
nonzero returns indicate system-dependent information; e.g. in
VMS isdir("[.FOO]") returns 1 but isdir("FOO.DIR;1") returns 2
to indicate the directory-file name is in a format that needs
conversion before it can be combined with a filename. Returns:
0: not a directory (including any kind of error)
1: it is an existing directory
char *
zfcdat(name) char *name;
Returns modification (preferably, otherwise creation) date/time
of file whose name is given in the argument string. Return
value is a pointer to a string of the form yyyymmdd hh:mm:ss,
for example 19931231 23:59:59, which represents the local time
(no timezone or daylight savings time finagling required).
Returns the null string ("") on failure. The text pointed to by
the string pointer might be in a static buffer, and so should
be copied to a safe place by the caller before any subsequent
calls to this function.
struct zfnfp *
zfnqfp(fn, buflen, buf) char * fn; int buflen; char * buf;
Given the filename fn, the corresponding fully qualified,
absolute filename is placed into the buffer buf, whose length
is buflen. On failure returns a NULL pointer. On success
returns a pointer to a struct zfnfp containing pointers to the
full pathname and to just the filename, and an int giving the
length of the full pathname. All references to this function in
mainline code must be protected by #ifdef ZFNQFP..#endif,
because it is not present in all of the ck*fio.c modules. So if
you implement this function in a version that did not have it
before, be sure to add #define ZFNQFP in the appropriate spot
in ckcdeb.h or in the build-procedure CFLAGS.
zcmpfn(s1,s2) char * s2, * s2;
Compares two filenames to see if they refer to the same.
Internally, the arguments can be converted to fully qualified
pathnames, e.g. with zfnqfp(), realpath(), or somesuch. In Unix
or other systems where symbolic links exist, the link should be
resolved before making the comparison or looking at the inodes.
0: Files are not identical.
1: Files are identical.
zfseek(pos) long pos;
Positions the input pointer on the current input file to the
given position. The pos argument is 0-based, the offset
(distance in bytes) from beginning of the file. Needed for
RESEND, PSEND, and other recovery operations. This function is
not necessarily possible on all systems, e.g. record-oriented
systems. It should only be used on binary files (i.e. files we
are sending in binary mode) and stream-oriented file systems.
-1: on failure.
0: On success.
zchdir(dirnam) char *dirnam;
Changes current or default directory to the one given in
dirnam. Returns:
0: On failure.
1: on success.
zchki(fn) char *fn;
Check to see if file with name fn is a regular, readable,
existing file, suitable for Kermit to send -- not a directory,
not a symbolic link, etc. Returns:
-3: if file exists but is not accessible (e.g.
-2: if file exists but is not of a readable type (e.g. a
-1: on error (e.g. file does not exist, or fn is garbage);
>=0: (length of file) if file exists and is readable.
Also see isdir(), zgetfs().
zchkpid(pid) unsigned long pid;
1: If the given process ID (e.g. pid in UNIX) is valid and
0: otherwise.
zgetfs(fn) char *fn;
Gets the size of the given file, regardless of accessibility.
Used for directory listings. Unlike zchki(), should return the
size of any kind of file, even a directory. zgetfs() also
should serve as a mini "get file info" function that can be
used until we design a better one, by also setting some global
int zgfs_link = 1/0 = file is (not) a symbolic link.
int zgfs_dir = 1/0 = file is (not) a directory.
char linkname[] = if zgfs_link != 0, name of file link points
-1: on error (e.g. file does not exist, or fn is garbage);
>=0: (length of file) if file exists and is readable.
zchko(fn) char *fn;
Checks to see if a file of the given name can be created.
-1: if file cannot be created, or on any kind of error.
0: if file can be created.
zchkspa(fn,len) char *f; long len;
Checks to see if there is sufficient space to store the file
named fn, which is len bytes long. If you can't write a
function to do this, then just make a dummy that always returns
1; higher level code will recover from disk-full errors. The
receiving Kermit uses this function to refuse an incoming file
based on its size, via the attribute mechanism. Returns:
-1: on error.
0: if there is not enough space.
1: if there is enough space.
zchin(n,c) int n; int *c;
Gets a character from file number n, return it in c (call with
&c). Returns:
-1: on failure, including EOF.
0: on success with character in c.
zchout(n,c) int n; char c;
Writes the character c to file number n. Returns:
-1: on error.
0: on success.
zclose(n) int n;
Closes file number n. Returns:
-1: on error.
1: on success.
zdelet(fn) char *name;
Attempts to delete (remove, erase) the named file. Returns:
-1: on error.
1: if file was deleted successfully.
char *
zgperm(char * f)
Returns a pointer to the system-dependent numeric
permissions/protection string for file f, or NULL upon failure.
Used if CK_PERMS is defined.
char *
ziperm(char * f)
Returns a pointer to the system-dependent symbolic
permissions/protection string for file f, or NULL upon failure.
Used if CK_PERMS is defined. Example: In UNIX zgperm(f) might
return "100770", but ziperm() might return "-rwxrwx---". In
VMS, zgperm() would return a hexadecimal string, but ziperm()
would return something like "(RWED,RWED,RE,)".
char *
Returns a pointer to the name of the current directory, folder,
etc, or a NULL pointer if the current directory cannot be
determined. If possible, the directory specification should be
(a) fully specified, e.g. as a complete pathname, and (b) be
suitable for appending a filename. Thus, for example, Unix
directory names should end with '/'. VMS directory names should
look like DEV:[NAME] (rather than, say, NAME.DIR;1).
char *
Returns a pointer to a string containing the user's home
directory, or NULL upon error. Should be formatted like
zgtdir() (q.v.).
Fill buffer from input file. This function is used by the macro
zminchar(), which is defined in ckcker.h. zminchar() manages
its own buffer, and calls zinfill() to fill it whenever it
becomes empty. It is used only for sending files, and reads
characters only from file number ZIFILE. zinfill() returns -1
upon end of file, -2 upon fatal error, and -3 upon timeout
(e.g. when reading from a pipe); otherwise it returns the first
character from the buffer it just read.
Kills the current job, session, process, etc, logs out,
disappears. Used by the Kermit server when it receives a BYE
command. On failure, returns -1. On success, does not return at
all! This function should not be called until all other steps
have been taken to close files, etc.
zstrip(fn,&fn2) char *fn1, **fn2;
Strips device and directory, etc, from file specification fn,
leaving only the filename (including "extension" or "filetype"
-- the part after the dot). For example DUA0:[PROGRAMS]OOFA.C;3
becomes OOFA.C, or /usr/fdc/oofa.c becomes oofa.c. Returns a
pointer to result in fn2.
zsetperm(char * file, unsigned int code)
Set permissions of file to given system-dependent code. 0: On
1: on success.
zsetroot(char * dir)
Sets the root for the user's file access, like Unix chroot(),
but does not require privilege. In Unix, this must be
implemented entirely by Kermit's own file access routines.
1: Success
-1: Invalid argument
-3: Internal error
-4: Access to given directory denied
-5: New root not within old root
zinroot(char * file)
If no root is set (zsetroot()), returns 1.
Otherwise, if given file is in the root, returns 1.
Otherwise, returns 0.
zltor(fn,fn2) char *fn1, *fn2;
Local-To-Remote filename translation. OBSOLETE: replaced by
nzltor() (q.v.). Translates the local filename fn into a format
suitable for transmission to an arbitrary type of computer, and
copies the result into the buffer pointed to by fn2.
Translation may involve (a) stripping the device and/or
directory/path name, (b) converting lowercase to uppercase, (c)
removing spaces and strange characters, or converting them to
some innocuous alphabetic character like X, (d) discarding or
converting extra periods (there should not be more than one).
Does its best. Returns no value. name2 is a pointer to a
buffer, furnished by the caller, into which zltor() writes the
resulting name. No length checking is done.
#ifdef NZLTOR
nzltor(fn,fn2,convert,pathnames,max) char *fn1,*fn2; int
Replaces zltor(). This new version handles pathnames and checks
length. fn1 and fn2 are as in zltor(). This version is called
unconditionally for each file, rather than only when filename
conversion is enabled. Pathnames can have the following values:
PATH_OFF: Pathname, if any, is to be stripped
PATH_REL: The relative pathname is to be included
PATH_ABS: The full pathname is to be included
After handling pathnames, conversion is done to the result as
in the zltor() description if convert != 0; if relative or
absolute pathnames are included, they are converted to UNIX
format, i.e. with slash (/) as the directory separator. The max
parameter specifies the maximum size of fn2. If convert > 0,
the regular conversions are done; if convert < 0, minimal
conversions are done (we skip uppercasing the letters, we allow
more than one period, etc; this can be used when we know our
partner is UNIX or similar).
#endif /* NZLTOR */
nzxpand(fn,flags) char *fn; int flags;
Replaces zxpand(), which is obsolete as of C-Kermit 7.0.
Call with:
fn = Pointer to filename or pattern.
flags = option bits:
flags & ZX_FILONLY Match regular files
flags & ZX_DIRONLY Match directories
flags & ZX_RECURSE Descend through directory tree
flags & ZX_MATCHDOT Match "dot files"
flags & ZX_NOBACKUP Don't match "backup files"
flags & ZX_NOLINKS Don't follow symlinks.
Returns the number of files that match fn, with data structures
set up so the first file (if any) will be returned by the next
znext() call. If ZX_FILONLY and ZX_DIRONLY are both set, or
neither one is set, files and directories are matched. Notes:
1. It is essential that the number returned by nzxpand() reflect
the actual number of filenames that will be returned by
znext() calls. In other words:
for (n = nzxpand(string,flags); n > 0; n--) {
printf("%s\n", buf);
should print all the file names; no more, no less.
2. In UNIX, DOS, OS-9, etc, where directories contain entries
for themselves (.) and the superior directory (..), these
should NOT be included in the list under any circumstances,
including when ZX_MATCHDOT is set.
3. Additional option bits might be added in the future, e.g. for
sorting (sort by date/name/size, reverse/ascending, etc).
Currently this is done only in higher level code (through a
hack in which the nzxpand() exports its filename array, which
is not portable because not all OS's can use this mechanism).
zmail(addr,fn) char *addr, fn;
Send the local, existing file fn as e-mail to the address addr.
0: on success
2: if mail delivered but temp file can't be deleted
-2: if mail can't be delivered
zmkdir(path) char *path;
The path can be a file specification that might contain
directory information, in which the filename is expected to be
included, or an unambiguous directory specification (e.g. in
UNIX it must end with "/"). This routine attempts to create any
directories in the given path that don't already exist. Returns
0 or greater success: no directories needed creation, or else
all directories that needed creation were created successfully;
the return code is the number of directories that were created.
Returns -1 on failure to create any of the needed directories.
zrmdir(path) char *path;
Attempts to remove the given directory. Returns 0 on success,
-1 on failure. The detailed semantics are open -- should it
fail if the directory contains any files or subdirectories,
etc. It is probably best for this routine to behave in whatever
manner is customary on the underlying platform; e.g. in UNIX,
VMS, DOS, etc, where directories can not be removed unless they
are empty.
znewn(fn,s) char *fn, **s;
Transforms the name fn into a filename that is guaranteed to be
unique. If the file fn does not exist, then the new name is the
same as fn; Otherwise, it's different. this function does its
best, returns no value. New name is created in caller's space.
Call like this: znewn(old,&new);. The second parameter is a
pointer to the new name. This pointer is set by znewn() to
point to a static string in its own space, so be sure to the
result to a safe place before calling this function again.
znext(fn) char *fn;
Copies the next file name from a file list created by zxpand()
into the string pointed to by fn (see zxpand). If no more
files, then the null string is placed there. Returns 0 if there
are no more filenames, with 0th element the array pointed to by
fn set to NUL. If there is a filename, it is stored in the
array pointed to by fn and a positive number is returned. NOTE:
This is a change from earlier definitions of this function
(pre-1999), which returned the number of files remaining; thus
0 was the return value when returning the final file. However,
no mainline code ever depended on the return value, so this
change should be safe.
zopeni(n,fn) int n; char *fn;
Opens the file named fn for input as file number n. Returns:
0: on failure.
1: on success.
zopeno(n,fn,zz,fcb) int n; char *name; struct zattr *zz; struct
filinfo *fcb;
Attempts to open the named file for output as file number n. zz
is a Kermit file attribute structure as defined in ckcdeb.h,
containing various information about the file, including its
size, creation date, and so forth. This function should attempt
to honor as many of these as possible. fcb is a "file control
block" in the traditional sense, defined in ckcdeb.h,
containing information relevant to complicated file systems
like VMS (RMS), IBM MVS, etc, like blocksize, record length,
organization, record format, carriage control, etc. Returns:
0: on failure.
1: on success.
Dumps a file output buffer. Used with the macro zmchout()
defined in ckcker.h. Used only with file number ZOFILE, i.e.
the file that is being received by Kermit during file transfer.
-1: on failure.
0: on success.
zprint(p,fn) char *p, *f;
Prints the file with name fn on a local printer, with options
p. Returns:
0: on success
3: if file sent to printer but can't be deleted
-3: if file can't be printed
zrename(fn,fn2) char *fn, *fn2;
Changes the name of file fn to fn2. If fn2 is the name of an
existing directory, or a file-structured device, then file fn
is moved to that directory or device, keeping its original
name. If fn2 lacks a directory separator when passed to this
function, an appropriate one is supplied. Returns:
-1: on failure.
0: on success.
zcopy(source,dest) char * source, * dest;
Copies the source file to the destination. One file only. No
wildcards. The destination string may be a filename or a
directory name. Returns:
0: on success.
<0: on failure:
-2: source file is not a regular file.
-3: source file not found.
-4: permission denied.
-5: source and destination are the same file.
-6: i/o error.
-1: other error.
char *
zlocaltime(char *)
Call with: "yyyymmdd hh:mm:ss" GMT/UTC date-time. Returns
pointer to local date-time string "yyyymmdd hh:mm:ss" on
success, NULL on failure.
zrtol(fn,fn2) char *fn, *fn2;
Remote-To-Local filename translation. OBSOLETE: replaced by
nzrtol(). Translates a "standard" filename to a local filename.
For example, in Unix this function might convert an
all-uppercase name to lowercase, but leave lower- or mix-case
names alone. Does its best, returns no value. New name is in
string pointed to by fn2. No length checking is done.
#ifdef NZLTOR
nzrtol(fn,fn2,convert,pathnames,max) char *fn1,*fn2; int
Replaces zrtol. Like zrtol but handles pathnames and checks
length. See nzltor for detailed description of parameters.
#endif /* NZLTOR */
zsattr(xx) struct zattr *xx;
Fills in a Kermit file attribute structure for the file which
is to be sent, namely the currently open ZIFILE. Note that this
is not a very good design, but we're stuck with it. Callers
must ensure that zsattr() is called only on real files, not on
pipes, internally generated file-like objects such as server
REMOTE command responses, etc. Returns:
-1: on failure.
0: on success with the structure filled in.
If any string member is null, it should be ignored by the
If any numeric member is -1, it should be ignored by the
zshcmd(s) char *s;
s contains to pointer to a command to be executed by the host
computer's shell, command parser, or operating system. If the
system allows the user to choose from a variety of command
processors (shells), then this function should employ the
user's preferred shell. If possible, the user's job
(environment, process, etc) should be set up to catch keyboard
interruption signals to allow the user to halt the system
command and return to Kermit. The command must run in ordinary,
unprivileged user mode. If possible, this function should
return -1 on failure to start the command, or else it should
return 1 if the command succeeded and 0 if it failed.
zshcmd() and zsyscmd() should set this to the command's actual
exit status code if possible.
zsyscmd(s) char *s;
s contains to pointer to a command to be executed by the host
computer's shell, command parser, or operating system. If the
system allows the user to choose from a variety of command
processors (shells), then this function should employ the
system standard shell (e.g. /bin/sh for Unix), so that the
results will always be the same for everybody. If possible, the
user's job (environment, process, etc) should be set up to
catch keyboard interruption signals to allow the user to halt
the system command and return to Kermit. The command must run
in ordinary, unprivileged user mode. If possible, this function
should return -1 on failure to start the command, or else it
should return 1 if the command succeeded and 0 if it failed.
z_exec(s,args) char * s; char * args[];
This one executes the command s (which is searched for using
the system's normal searching mechanism, such as PATH in UNIX),
with the given argument vector, which follows the conventions
of UNIX argv[]: the name of the command pointed to by element
0, the first arg by element 1, and so on. A null args[] pointer
indicates the end of the arugment list. All open files must
remain open so the exec'd process can use them. Returns only if
zsinl(n,s,x) int n, x; char *s;
Reads a line from file number n. Writes the line into the
address s provided by the caller. Writing terminates when
newline is read, but with newline discarded. Writing also
terminates upon EOF or if length x is exhausted. Returns:
-1: on EOF or error.
0: on success.
zsout(n,s) int n; char *s;
Writes the string s out to file number n. Returns:
-1: on failure.
0: on success.
zsoutl(n,s) int n; char *s;
Writes the string s out to file number n and adds a line
(record) terminator (boundary) appropriate for the system and
the file format. Returns:
-1: on failure.
0: on success.
zsoutx(n,s,x) int n, x; char *s;
Writes exactly x characters from string s to file number n. If
s has fewer than x characters, then the entire string s is
written. Returns:
-1: on failure.
>= 0: on success, the number of characters actually written.
zstime(fn,yy,x) char *fn; struct zattr *yy; int x;
Sets the creation date (and other attributes) of an existing
file, or compares a file's creation date with a given date.
Call with:
fn: pointer to name of existing file.
yy: Pointer to a Kermit file attribute structure in which yy->date.val
is a date of the form yyyymmdd hh:mm:ss, e.g. 19900208 13:00:00, which
is to be used for setting or comparing the file date. Other attributes
in the struct can also be set, such as the protection/permission (See
[110]Appendix I), when it makes sense (e.g. "yy->lprotect.val" can be
set if the remote system ID matches the local one).
x: A function code: 0 means to set the file's creation date as given.
1 means compare the date from the yy struct with the file's date.
-1: on any kind of error.
0: if x is 0 and the file date was set successfully.
0: if x is 1 and date from attribute structure > file
creation date.
1: if x is 1 and date from attribute structure <= file
creation date.
zstrip(name,name2) char *name, **name2;
Strips pathname from filename "name". Constructs the resulting
string in a static buffer in its own space and returns a
pointer to it in name2. Also strips device name, file version
numbers, and other "non-name" material.
zxcmd(n,s) char *s;
Runs a system command so its output can be accessed as if it
were file n. The command is run in ordinary, unprivileged user
If n is ZSTDIO or ZCTERM, returns -1.
If n is ZIFILE or ZRFILE, then Kermit reads from the command,
otherwise Kermit writes to the command.
Returns 0 on error, 1 on success.
zxpand(fn) char *fn;
OBSOLETE: Replaced by nzxpand(), q.v.
Returns the number of files returned by the most recent
nzxpand() call, and resets the list to the beginning so the
next znext() call returns the first file. Returns -1 if zxpand
has not yet been called. If this function is available,
ZXREWIND should be defined; otherwise it should not be
#endif /* ZXREWIND */
xsystem(cmd) char *cmd;
Executes the system command without redirecting any of its i/o,
similar (well, identical) to system() in Unix. But before
passing the command to the system, xsystem() ensures that all
privileges are turned off, so that the system command executes
in ordinary unprivileged user mode. If possible, xsystem()
returns the return code of the command that was executed.
4.E.2.2. IKSD Variables and Functions
These must be implemented in any C-Kermit version that is to be
installed as an Internet Kermit Service Daemon (IKSD). IKSD is
expected to be started by the Internet Daemon (e.g. inetd) with its
standard i/o redirected to the incoming connection.
int ckxanon;
Nonzero if anonymous logins allowed.
extern int inserver;
Nonzero if started in IKSD mode.
extern int isguest;
Nonzero if IKSD and user logged in anonymously.
extern char * homdir;
Pointer to user's home directory.
extern char * anonroot;
Pointer to file-system root for anonymous users.
Existing functions must make "if (inserver && isguest)" checks for
actions that would not be legal for guests: zdelete(), zrmdir(),
zprint(), zmail(), etc.
zvuser(name) char * name;
Verifies that user "name" exists and is allowed to log in. If
the name is "ftp" or "anonymous" and ckxanon != 0, a guest
login is set up. Returns 0 if user not allowed to log in,
nonzero if user may log in.
zvpass(string) char * string;
Verifies password of the user from the most recent zvuser()
call. Returns nonzero if password is valid for user, 0 if it
isn't. Makes any appropriate system log entries (IKSD logins,
failed login attempts, etc). If password is valid, logs the
user in as herself (if real user), or sets up restricted
anonymous access if user is guest (e.g. changes file-system
root to anonroot and sets isguest = 1).
Begins any desired system logging of an IKSD session.
Terminates an IKSD session. In most cases this is simply a
wrapper for exit() or doexit(), with some system logging added.
4.E.2.3. Privilege Functions
These functions are used by C-Kermit to adapt itself to operating
systems where the program can be made to run in a "privileged" mode,
e.g. setuid or setgid in Unix. C-Kermit should NOT read and write
files or start subprocesses as a privileged program. This would
present a serious threat to system security. The security package has
been installed to prevent such security breaches by turning off the
program's special privileges at all times except when they are needed.
In UNIX, the only need Kermit has for privileged status is access to
the UUCP lockfile directory, in order to read, create, and destroy
lockfiles, and to open communication devices that are normally
protected against the user (see the [111]Unix C-Kermit Installation
Instructions for discussion). Therefore, privileges should only be
enabled for these operations and disabled at all other times. This
relieves the programmer of the responsibility of putting expensive and
unreliable access checks around every file access and subprocess
Strictly speaking, these functions are not required in all C-Kermit
implementations, because their use (so far, at least) is internal to
the Group E modules. However, they should be included in all C-Kermit
implementations for operating systems that support the notion of a
privileged program (UNIX, RSTS/E, what others?).
Determine whether the program is running in privileged status.
If so, turn off the privileges, in such a way that they can be
turned on again when needed. Called from sysinit() at program
startup time. Returns:
0 on success
nonzero on failure, in which case the program should halt
If the program is not privileged, this function does nothing.
If the program is privileged, this function returns it to
privileged status. priv_ini() must have been called first.
0 on success
nonzero on failure
Turns privileges off (if they are on) in such a way that they
can be turned back on again. Returns:
0 on success
nonzero on failure
Turns privileges off in such a way that they cannot be turned
back on. Returns:
0 on success
nonzero on failure
Attempts to turns privileges off in such a way that they can be
turned on again later. Then checks to make sure that they were
really turned off. If they were not really turned off, then
they are cancelled permanently. Returns:
0 on success
nonzero on failure
4.E.2.4. Console-Related Functions
These relate to the program's "console", or controlling terminal, i.e.
the terminal that the user is logged in on and types commands at, or
on a PC or workstation, the actual keyboard and screen.
conbin(esc) char esc;
Puts the console into "binary" mode, so that Kermit's command
parser can control echoing and other treatment of characters
that the user types. esc is the character that will be used to
get Kermit's attention during packet mode; puts this in a
global place. Sets the ckxech variable. Returns:
-1: on error.
0: on success.
concb(esc) char esc;
Put console in "cbreak" (single-character wakeup) mode. That
is, ensure that each console character is available to the
program immediately when the user types it. Otherwise just like
conbin(). Returns:
-1: on error.
0: on success.
Returns a number, 0 or greater, the number of characters
waiting to be read from the console, i.e. the number of
characters that the user has typed that have not been read yet
by Kermit.
Returns the speed ("baud rate") of the controlling terminal, if
known, otherwise -1L.
congks(timo) int timo;
Get Keyboard Scancode. Reads a keyboard scan code from the
physical console keyboard. If the timo parameter is greater
than zero, then times out and returns -2 if no character
appears within the given number of seconds. Upon any other kind
of error, returns -1. Upon success returns a scan code, which
may be any positive integer. For situations where scan codes
cannot be read (for example, when an ASCII terminal is used as
the job's controlling terminal), this function is identical to
coninc(), i.e. it returns an 8-bit character value. congks() is
for use with workstations whose keyboards have Alternate,
Command, Option, and similar modifier keys, and Function keys
that generate codes greater than 255.
Console get modes. Gets the current console terminal modes and
saves them so that conres() can restore them later. Returns 1
if it got the modes OK, 0 if it did nothing (e.g. because
Kermit is not connected with any terminal), -1 on error.
coninc(timo) int timo;
Console Input Character. Reads a character from the console. If
the timo parameter is greater than zero, then coninc() times
out and returns -2 if no character appears within the given
number of seconds. Upon any other kind of error, returns -1.
Upon success, returns the character itself, with a value in the
range 0-255 decimal.
conint(f,s) SIGTYP (*f)(), (*s)();
Sets the console to generate an interrupt if the user types a
keyboard interrupt character, and to transfer control the
signal-handling function f. For systems with job control, s is
the address of the function that suspends the job. Sets the
global variable "backgrd" to zero if Kermit is running in the
foreground, and to nonzero if Kermit is running in the
background. See ckcdeb.h for the definition of SIGTYP. No
return value.
Console no interrupts. Disable keyboard interrupts on the
console. No return value.
conoc(c) char c;
Writes character c to the console terminal. Returns:
0 on failure, 1 on success.
conol(s) char *s;
Writes string s to the console. Returns -1 on error, 0 or
greater on success.
conola(s) char *s[]; {
Writes an array of strings to the console. Returns -1 on error,
0 or greater on success.
conoll(s) char *s;
Writes string s to the console, followed by the necessary line
termination characters to put the console cursor at the
beginning of the next line. Returns -1 on error, 0 or greater
on success.
Restores the console terminal to the modes obtained by congm().
Returns: -1 on error, 0 on success.
conxo(x,s) int x; char *s;
Write x characters from string s to the console. Returns 0 or
greater on success, -1 on error.
char *
Returns a pointer to the designator of the console keyboard
type. For example, on a PC, this function would return "88",
"101", etc. Upon failure, returns a pointer to the empty
4.E.2.5. Communications Functions
The communication device is the device used for terminal emulation and
file transfer. It may or may not be the same device as the console,
and it may or may not be a terminal (serial-port) device; it could
also be a network connection. For brevity, the communication device is
referred to here as the "tty". When the communication device is the
same as the console device, Kermit is said to be in remote mode. When
the two devices are different, Kermit is in local mode.
Returns the number of characters that have arrived at the
communication device but have not yet been read by ttinc(),
ttinl(), and friends. If communication input is buffered (and
it should be), this is the sum of the number of unread
characters in Kermit's buffer PLUS the number of unread
characters in the operating system's internal buffer. The call
must be nondestructive and nonblocking, and as inexpensive as
possible. Returns:
0: or greater on success,
0: in case of internal error,
-1: or less when it determines the connection has been broken,
or there is no connection.
That is, a negative return from ttchk() should reliably
indicate that there is no usable connection. Furthermore,
ttchk() should be callable at any time to see if the connection
is open. When the connection is open, every effort must be made
to ensure that ttchk returns an accurate number of characters
waiting to be read, rather than just 0 (no characters) or 1 (1
or more characters), as would be the case when we use select().
This aspect of ttchk's operation is critical to successful
operation of sliding windows and streaming, but "nondestructive
buffer peeking" is an obscure operating system feature, and so
when it is not available, we have to do it ourselves by
managing our own internal buffer at a level below ttinc(),
ttinl(), etc, as in the UNIX version (non-FIONREAD case).
An external global variable, clsondisc, if nonzero, means that
if a serial connection drops (carrier on-to-off transition
detected by ttchk()), the device should be closed and released
Closes the communication device (tty or network). If there were
any kind of exclusive access locks connected with the tty,
these are released. If the tty has a modem connection, it is
hung up. For true tty devices, the original tty device modes
are restored. Returns:
-1: on failure.
0: on success.
Flush communications input buffer. If any characters have
arrived but have not yet been read, discard these characters.
If communications input is buffered by Kermit (and it should
be), this function flushes Kermit's buffer as well as the
operating system's internal input buffer. Returns:
-1: on failure.
0: on success.
Flush tty output buffer. If any characters have been written
but not actually transmitted (e.g. because the system has been
flow-controlled), remove them from the system's output buffer.
(Note, this function is not actually used, but it is
recommended that all C-Kermit programmers add it for future
use, even if it is only a dummy function that returns 0
Looks for the modem signals CTS, DSR, and CTS, and returns
those that are on in as its return value, in a bit mask as
described for ttwmdm, in which a bit is on (1) or off (0)
according to whether the corresponding signal is on (asserted)
or off (not asserted). Return values:
-3: Not implemented
-2: if the line does not have modem control
-1: on error
>=0: on success, with bit mask containing the modem signals.
Returns the current tty speed in BITS (not CHARACTERS) per
second, or -1 if it is not known or if the tty is really a
network, or upon any kind of error. On success, the speed
returned is the actual number of bits per second, like 1200,
9600, 19200, etc.
Get terminal window size. Returns -1 on error, 0 if the window
size can't be obtained, 1 if the window size has been
successfully obtained. Upon success, the external global
variables tt_rows and tt_cols are set to the number of screen
rows and number of screen columns, respectively. As this
function is not implemented in all ck*tio.c modules, calls to
it must be wrapped in #ifdef CK_TTGWSIZ..#endif. NOTE: This
function must be available to use the TELNET NAWS feature
(Negotiate About Window Size) as well as Rlogin.
Hang up the current tty device. For real tty devices, turn off
DTR for about 1/3-1/2 second (or other length of time,
depending on the system). If the tty is really a network
connection, close it. Returns:
-1: on failure.
0: if it does not even try to hang up.
1: if it believes it hung up successfully.
Turns off all pending timer interrupts.
ttinc(timo) int timo; (function is old, return codes are new)
Reads one character from the communication device. If timo is
greater than zero, wait the given number of seconds and then
time out if no character arrives, otherwise wait forever for a
character. Returns:
-3: internal error (e.g. tty modes set wrong)
-2: communications disconnect
-1: timeout or other error
>=0: the character that was read.
It is HIGHLY RECOMMENDED that ttinc() be internally buffered so
that calls to it are relatively inexpensive. If it is possible
to to implement ttinc() as a macro, all the better, for example
something like:
#define ttinc(t) ( (--txbufn >= 0) ? txbuf[ttbufp++] : txbufr(t) )
(see description of txbufr() below)
ttinl(dest,max,timo,eol,start,turn) int max,timo,turn; CHAR
*dest, eol, start;
ttinl() is Kermit's packet reader. Reads a packet from the
communications device, or up to max characters, whichever
occurs first. A line is a string of characters starting with
the start character up to and including the character given in
eol or until the length is exhausted, or, if turn != 0, until
the line turnaround character (turn) is read. If turn is 0,
ttinl() *should* use the packet length field to detect the end,
to allow for the possibility that the eol character appears
unprefixed in the packet data. (The turnaround character is for
half-duplex linemode connections.)
If timo is greater than zero, ttinl() times out if the eol
character is not encountered within the given number of seconds
and returns -1.
The characters that were input are copied into "dest" with
their parity bits stripped if parity is not none. The first
character copied into dest should be the start character, and
the last should be the final character of the packet (the last
block check character). ttinl() should also absorb and discard
the eol and turn characters, and any other characters that are
waiting to be read, up until the next start character, so that
subsequent calls to ttchk() will not succeed simply because
there are some terminators still sitting in the buffer that
ttinl() didn't read. This operation, if performed, MUST NOT
BLOCK (so if it can't be performed in a guaranteed nonblocking
way, don't do it).
On success, ttinl() returns the number of characters read.
Optionally, ttinl() can sense the parity of incoming packets.
If it does this, then it should set the global variable ttprty
accordingly. ttinl() should be coded to be as efficient as
possible, since it is at the "inner loop" of packet reception.
ttinl() returns:
-1: Timeout or other possibly correctable error.
-2: Interrupted from keyboard.
-3: Uncorrectable i/o error -- connection lost, configuration
problem, etc.
>=0: on success, the number of characters that were actually
read and placed in the dest buffer, not counting the trailing
ttoc(c) char c;
Outputs the character c to the communication line. If the
operation fails to complete within two seconds, this function
returns -1. Otherwise it returns the number of characters
actually written to the tty (0 or 1). This function should only
be used for interactive, character-mode operations, like
terminal connection, script execution, dialer i/o, where the
overhead of the signals and alarms does not create a
operation fails within a "certain amount of time"... which
might be dependent on the communication method, speed, etc. In
particular, flow-control deadlocks must be accounted for and
broken out of to prevent the program from hanging indefinitely,
ttol(s,n) int n; char *s;
Kermit's packet writer. Writes the n characters of the string
pointed to to by s. NOTE: It is ttol's responsibility to write
ALL of the characters, not just some of them. Returns:
-1: on a possibly correctable error (so it can be retried).
-3: on a fatal error, e.g. connection lost.
>=0: on success, the actual number of characters written (the
specific number is not actually used for anything).
ttopen(ttname,lcl,modem,timo) char *ttname; int *lcl, modem,
Opens a tty device, if it is not already open. ttopen must
check to make sure the SAME device is not already open; if it
is, ttopen returns successfully without doing anything. If a
DIFFERENT device is currently open, ttopen() must call ttclos()
to close it before opening the new one.
character string - device name or network host
If called with lcl < 0, sets value of lcl as
0: the terminal named by ttname is the job's
controlling terminal.
1: the terminal named by ttname is not the job's
controlling terminal.
If the device is already open, or if the requested
device can't be opened, then lcl remains (and is
returned as) -1.
Less than zero: this is the negative of the network
type, and ttname is a network host name. Network
types (from [112]ckcnet.h:
NET_TCPB 1 TCP/IP Berkeley (socket) (implemented in [113]ckutio.c)
NET_TCPA 2 TCP/IP AT&T (streams) (not yet implemented)
NET_DEC 3 DECnet (not yet implemented)
Zero or greater: ttname is a terminal device name.
Zero means a direct connection (don't use modem
signals). Positive means use modem signals
depending on the current setting of ttcarr (see
> 0: number of seconds to wait for open() to return
before timing out.
<=0: no timer, wait forever (e.g. for incoming
For real tty devices, ttopen() attempts to gain
exclusive access to the tty device, for example in
UNIX by creating a "lockfile" (in other operating
systems, like VMS, exclusive access probably
requires no special action).
Side effects:
Copies its arguments and the tty file descriptor to
global variables that are available to the other
tty-related functions, with the lcl value altered as
described above. Gets all parameters and settings
associated with the line and puts them in a global area,
so that they can be restored by ttres(), e.g. when the
device is closed.
0: on success
-5: if device is in use
-4: if access to device is denied
-3: if access to lock mechanism denied
-2: upon timeout waiting for device to open
-1: on other error
ttpkt(speed,flow,parity) long speed; int flow, parity;
Puts the currently open tty device into the appropriate modes
for transmitting and receiving Kermit packets.
if speed > -1, and the device is a true tty device,
and Kermit is in local mode, ttpkt also sets the
if in the range 0-3, ttpkt selects the
corresponding type of flow control. Currently 0 is
defined as no flow control, 1 is Xon/Xoff, and no
other types are defined. If (and this is a horrible
hack, but it goes back many years and will be hard
to eradicate) flow is 4, then the appropriate tty
modes are set for modem dialing, a special case in
which we talk to a modem-controlled line without
requiring carrier. If flow is 5, then we require
This is simply copied into a global variable so
that other functions (like ttinl, ttinc, etc) can
use it.
Side effects:
Copies its arguments to global variables, flushes the
terminal device input buffer.
-1: on error.
0: on success.
Enables the given type of flow control on the open serial
communications device immediately. Arguments are the FLO_xxx
values from ckcdeb.h, except FLO_DIAL, FLO_DIAX, or FLO_AUTO,
which are not actual flow-control types. Returns 0 on success,
-1 on failure.
long *
Returns a pointer to an array of longs, or NULL on failure. On
success, element 0 of the array contains number, n, indicating
how many follow. Elements 1-n are serial speeds, expressed in
bits per second, that are legal on this platform. The user
interface may use this list to construct a menu, keyword table,
#endif /* TTSPDLIST */
Restores the tty device to the modes and settings that were in
effect at the time it was opened (see ttopen). Returns:
-1: on error.
0: on success.
ttruncmd(string) char * string;
Runs the given command on the local system, but redirects its
input and output to the communication (SET LINE, SET PORT, or
SET HOST) device. Returns:
0: on failure.
1: on success.
ttscarr(carrier) int carrier;
Copies its argument to a variable that is global to the other
tty-related functions, and then returns it. The values for
carrier are defined in ckcdeb.h: CAR_ON, CAR_OFF, CAR_AUTO.
ttopen(), ttpkt(), and ttvt() use this variable when deciding
how to open the tty device and what modes to select. The
meanings are these:
CAR_OFF: Ignore carrier at all times.
CAR_ON: Require carrier at all times, except when dialing. This means,
for example, that ttopen() could hang forever waiting for carrier if
it is not present.
CAR_AUTO: If the modem type is zero (i.e. the connection is direct),
this is the same as CAR_OFF. If the modem type is positive, then heed
carrier during CONNECT (ttvt mode), but ignore it at other times
(packet mode, during SET LINE, etc). Compatible with pre-5A versions
of C-Kermit. This should be the default carrier mode.
Kermit's DIAL command ignores the carrier setting, but
ttopen(), ttvt(), and ttpkt() all honor the carrier option in
effect at the time they are called. None of this applies to
remote mode (the tty device is the job's controlling terminal)
or to network host connections (modem type is negative).
Sends a BREAK signal on the tty device. On a real tty device,
send a real BREAK lasting approximately 275 milliseconds. If
this is not possible, simulate a BREAK by (for example)
dropping down some very low baud rate, like 50, and sending a
bunch of null characters. On a network connection, do the
appropriate network protocol for BREAK. Returns:
-1: on error.
0: on success.
Like ttsndb(), but sends a "Long BREAK" (approx 1.5 seconds).
For network connections, it is identical to ttsndb().
Currently, this function is used only if CK_LBRK is defined (as
it is for UNIX and VMS).
ttsspd(cps) int cps;
For serial devices only, set the device transmission speed to
(note carefully) TEN TIMES the argument. The argument is in
characters per second, but transmission speeds are in bits per
second. cps are used rather than bps because high speeds like
38400 are not expressible in a 16-bit int but longs cannot be
used because keyword-table values are ints and not longs. If
the argument is 7, then the bps is 75, not 70. If the argument
is 888, this is a special code for 75/1200 split-speed
operation (75 bps out, 1200 bps in). Returns:
-1: on error, meaning the requested speed is not valid or
>=0: on success (don't try to use this value for anything).
ttvt(speed,flow) long speed; int flow;
Puts the currently open tty device into the appropriate modes
for terminal emulation. The arguments are interpreted as in
ttpkt(). Side effects: ttvt() stores its arguments in global
variables, and sets a flag that it has been called so that
subsequent calls can be ignored so long as the arguments are
the same as in the last effective call. Other functions, such
as ttopen(), ttclose(), ttres(), ttvt(), etc, that change the
tty device in any way must unset this flag. In UNIX Kermit,
this flag is called tvtflg.
ttwmdm(mdmsig,timo) int mdmsig, timo;
Waits up to timo seconds for all of the given modem signals to
appear. mdmsig is a bit mask, in which a bit is on (1) or off
(0) according to whether the corresponding signal is to be
waited for. These symbols are defined in ckcdeb.h:
BM_CTS (bit 0) means wait for Clear To Send
BM_DSR (bit 1) means wait for Data Set Ready
BM_DCD (bit 2) means wait for Carrier Detect
-3: Not implemented.
-2: This line does not have modem control.
-1: Timeout: time limit exceeded before all signals were
1: Success.
ttxin(n,buf) int n; CHAR *buf;
Reads x characters from the tty device into the specified buf,
stripping parity if parity is not none. This call waits
forever, there is no timeout. This function is designed to be
called only when you know that at least x characters are
waiting to be read (as determined, for example, by ttchk()).
This function should use the same buffer as ttinc().
txbufr(timo) int timo;
Reads characters into the internal communications input buffer.
timo is a timeout interval, in seconds. 0 means no timeout,
wait forever. Called by ttinc() (and possibly ttxin() and
ttinl()) when the communications input buffer is empty. The
buffer should be called ttxbuf[], its length is defined by the
symbol TXBUFL. The global variable txbufn is the number of
characters available to be read from ttxbuf[], and txbufp is
the index of the next character to be read. Should not be
called if txbufn > 0, in which case the buffer does not need
refilling. This routine returns:
-2: Communications disconnect
-1: Timeout
>=0: A character (0 - 255) On success, the first character that
was read, with the variables txbufn and txbufp set
appropriately for any remaining characters.
NOTE: Currently this routine is used internally only by the
UNIX and VMS versions. The aim is to make it available to all
versions so there is one single coherent and efficient way of
reading from the communications device or network.
4.E.2.6. Miscellaneous system-dependent functions
ztime(s) char **s;
Returns a pointer, s, to the current date-and-time string in s.
This string must be in the fixed-field format associated with
the C runtime asctime() function, like: "Sun Sep 16 13:23:45
1973\n" so that callers of this function can extract the
different fields. The pointer value is filled in by ztime, and
the data it points to is not safe, so should be copied to a
safe place before use. ztime() has no return value. As a side
effect, this routine can also fill in the following two
external variables (which must be defined in the
system-dependendent modules for each platform):
long ztusec: Fraction of seconds of clock time, microseconds.
long ztmsec: Fraction of seconds of clock time, milliseconds.
If these variables are not set by zstime(), they remain at
their initial value of -1L.
Returns the current value of the elapsed time counter in
seconds (see rtimer), or 0 on any kind of error.
#ifdef GFTIMER
Returns the current value of the elapsed time counter in
seconds, as a floating point number, capable of representing
not only whole seconds, but also the fractional part, to the
millisecond or microsecond level, whatever precision is
available. Requires a function to get times at subsecond
precision, as well as floating-point support. That's why it's
#endif /* GFTIMER */
msleep(m) int m;
Sleeps (pauses, does nothing) for m milliseconds (a millisecond
is one thousandth of a second). Returns:
-1: on failure.
0: on success.
Sets the elapsed time counter to zero. If you want to time how
long an operation takes, call rtimer() when it starts and
gtimer when it ends. rtimer() has no return value.
#ifdef GFTIMER
Sets the elapsed time counter to zero. If you want to time how
long an operation takes, call rftimer() when it starts and
gftimer when it ends. rftimer() has no return value. Note:
rftimer() is to be used with gftimer() and rtimer() is to be
used with gtimer(). See the rftimer() description.
#endif /* GFTIMER */
Does whatever needs doing upon program start. In particular, if
the program is running in any kind of privileged mode, turns
off the privileges (see priv_ini()). Returns:
-1: on error.
0: on success.
Does whatever needs doing upon program exit. Returns:
-1: on error.
0: on success.
Suspends the Kermit process, puts it in the background so it
can be continued ("foregrounded") later. Returns:
-1: if this function is not supported.
0: on success.
[ [114]Contents ] [ [115]C-Kermit ] [ [116]Kermit Home ]
4.F. Group F: Network Support
As of version 5A, C-Kermit includes support for several networks.
Originally, this was just worked into the ttopen(), ttclos(), ttinc(),
ttinl(), and similar routines in [117]ckutio.c. But this made it
impossible to share this code with non-UNIX versions, like VMS,
AOS/VS, OS/2, etc. So as of edit 168, network code has been separated
out into its own module and header file, ckcnet.c and ckcnet.h:
[118]ckcnet.h: Network-related symbol definitions.
[119]ckcnet.c: Network i/o (TCP/IP, X.25, etc), shared by most
[120]cklnet.c: Network i/o (TCP/IP, X.25, etc) specific to Stratus
The routines and variables in these modules fall into two categories:
1. Support for specific network packages like SunLink X.25 and TGV
MultiNet, and:
2. support for specific network virtual terminal protocols like CCITT
X.3 and TCP/IP Telnet.
Category (1) functions are analogs to the tt*() functions, and have
names like netopen, netclos, nettinc, etc. Group A-D modules do not
(and must not) know anything about these functions -- they continue to
call the old Group E functions (ttopen, ttinc, etc). Category (2)
functions are protocol specific and have names prefixed by a protocol
identifier, like tn for telnet x25 for X.25.
ckcnet.h contains prototypes for all these functions, as well as
symbol definitions for network types, protocols, and network- and
protocol- specific symbols, as well as #includes for the header files
necessary for each network and protocol.
The following functions are to be provided for networks that do not
use normal system i/o (open, read, write, close):
To be called from within ttopen() when a network connection is
requested. Calling conventions and purpose same as Group E
To be called from within ttclos() when a network connection is
being closed. Calling conventions and purpose same as Group E
To be called from within ttchk(). Calling conventions and
purpose same as Group E ttchk().
To be called from within ttflui(). Calling conventions and
purpose same as Group E ttflui().
To send a network break (attention) signal. Calling conventions
and purpose same as Group E ttsndbrk().
To get a character from the network. Calling conventions same
as Group E ttsndbrk().
Send a "character" (byte) to the network. Calling conventions
same as Group E ttoc().
Send a "line" (sequence of bytes) to the network. Calling
conventions same as Group E ttol().
Conceivably, some systems support network connections simply by
letting you open a device of a certain name and letting you do i/o to
it. Others (like the Berkeley sockets TCP/IP library on UNIX) require
you to open the connection in a special way, but then do normal i/o
(read, write). In such a case, you would use netopen(), but you would
not use nettinc, nettoc, etc.
VMS TCP/IP products have their own set of functions for all network
operations, so in that case the full range of netxxx() functions is
The technique is to put a test in each corresponding ttxxx() function
to see if a network connection is active (or is being requested), test
for which kind of network it is, and if necessary route the call to
the corresponding netxxx() function. The netxxx() function must also
contain code to test for the network type, which is available via the
global variable ttnet.
[ [121]Contents ] [ [122]C-Kermit ] [ [123]Kermit Home ]
4.F.1. Telnet Protocol
(This section needs a great deal of updating...)
As of edit 195, Telnet protocol is split out into its own files, since
it can be implemented in remote mode, which does not have a network
[124]ckctel.h: Telnet protocol symbol definitions.
[125]ckctel.c: Telnet protocol.
The Telnet protocol is supported by the following variables and
int tn_init
Nonzero if telnet protocol initialized, zero otherwise.
Initialize the telnet protocol (send initial options).
Send a telnet option.
Receive and act on a telnet option from the remote.
Send terminal type using telnet protocol.
4.F.2. FTP Protocol
(To be filled in...)
4.F.3. HTTP Protocol
(To be filled in...)
4.F.4. X.25 Networks
These routines were written SunLink X.25 and have since been adapted
to at least on one other: IBM AIXLink/X.25.
Reads and prints X.25 diagnostics
X.25 out of band signal handler
Sends X.25 interrupt packet
Resets X.25 virtual circuit
Clear X.25 virtual circuit
X.25 status
Sets X.25 Q-bit
Resets X.25 Q-bit
Reads n characters from X.25 circuit.
Read a Kermit packet from X.25 circuit.
[ [126]Contents ] [ [127]C-Kermit ] [ [128]Kermit Home ]
4.F.5. Adding New Network Types
Example: Adding support for IBM X.25 and Hewlett Packard X.25. First,
add new network type symbols for each one. There are already some
network types defined for other X.25 packages:
NET_SX25 is the network-type ID for SunLink X.25.
NET_VX25 is the network-type ID for VOS X.25.
So first you should new symbols for the new network types, giving them
the next numbers in the sequence, e.g.:
#define NET_HX25 11 /* Hewlett-Packard X.25 */
#define NET_IX25 12 /* IBM X.25 */
This is in ckcnet.h.
Then we need symbols to say that we are actually compiling in the code
for these platforms. These would be defined on the cc command line:
-DIBMX25 (for IBM)
-DHPX25 (for HP)
So we can build C-Kermit versions for AIX and HP-UX both with and
without X.25 support (since not all AIX and IBM systems have the
needed libraries, and so an executable that was linked with them might
no load).
Then in ckcnet.h:
#ifdef IBMX25
#define ANYX25
#endif /* IBMX25 */
#ifdef HPX25
#define ANYX25
#endif /* HPX25 */
And then use ANYX25 for code that is common to all of them, and IBMX25
or HPX25 for code specific to IBM or HP.
It might also happen that some code can be shared between two or more
of these, but not the others. Suppose, for example, that you write
code that applies to both IBM and HP, but not Sun or VOS X.25. Then
you add the following definition to ckcnet.h:
#ifndef HPORIBMX25
#ifdef HPX25
#define HPORIBMX25
#ifdef IBMX25
#define HPORIBMX25
#endif /* IBMX25 */
#endif /* HPX25 */
#endif /* HPORIBMX25 */
You can NOT use constructions like "#if defined (HPX25 || IBMX25)";
they are not portable.
[ [129]Contents ] [ [130]C-Kermit ] [ [131]Kermit Home ]
4.G. Group G: Formatted Screen Support
So far, this is used only for the fullscreen local-mode file transfer
display. In the future, it might be extended to other uses. The
fullscreen display code is in and around the routine screenc() in
In the UNIX version, we use the curses library, plus one call from the
termcap library. In other versions (OS/2, VMS, etc) we insert dummy
routines that have the same names as curses routines. So far, there
are two methods for simulating curses routines:
1. In VMS, we use the Screen Management Library (SMG), and insert
stubs to convert curses calls into SMG calls.
2. In OS/2, we use the MYCURSES code, in which the stub routines
actually emit the appropriate escape sequences themselves.
Here are the stub routines:
tgetent(char *buf, char *term)
Arguments are ignored. Returns 1 if the user has a supported
terminal type, 0 otherwise. Sets a global variable (for
example, "isvt52" or "isdasher") to indicate the terminal type.
move(int row, int col)
Sends the escape sequence to position the cursor at the
indicated row and column. The numbers are 0-based, e.g. the
home position is 0,0.
Sends the escape sequence to clear the screen.
Sends the escape sequence to clear from the current cursor
position to the end of the line.
In the MYCURSES case, code must be added to each of the last three
routines to emit the appropriate escape sequences for a new terminal
clearok(curscr), wrefresh()
In real curses, these two calls are required to refresh the
screen, for example after it was fractured by a broadcast
message. These are useful only if the underlying screen
management service keeps a copy of the entire screen, as curses
and SMG do. C-Kermit does not do this itself.
[ [133]Contents ] [ [134]C-Kermit ] [ [135]Kermit Home ]
4.H. Group H: Pseudoterminal Support
(To be filled in...)
4.I. Group I: Security
(To be filled in...)
[ [136]Contents ] [ [137]C-Kermit ] [ [138]Kermit Home ]
I.1. Format of System-Dependent File Permissions in A-Packets
The format of this field (the "," attribute) is interpreted according
to the System ID ("." Attribute).
For UNIX (System ID = U1), it's the familiar 3-digit octal number, the
low-order 9 bits of the filemode: Owner, Group, World, e.g. 660 =
read/write access for owner and group, none for world, recorded as a
3-digit octal string. High-order UNIX permission bits are not
For VMS (System ID = D7), it's a 4-digit hex string, representing the
16-bit file protection WGOS fields (World,Group,Owner,System), in that
order (which is the reverse of how they're shown in a directory
listing); in each field, Bit 0 = Read, 1 = Write, 2 = Execute, 3 =
Delete. A bit value of 0 means permission is granted, 1 means
permission is denied. Sample:
r-02-01-^A]"A."D7""B8#119980101 18:14:05!#8531&872960,$A20B-!7(#512@ #.Y
A VMS directory listing shows the file's protection as (E,RWED,RED,RE)
which really means (S=E,O=RWED,G=RED,W=RE), which is reverse order
from the internal storage, so (RE,RED,RWED,E). Now translate each
letter to its corresponding bit:
RE=0101, RED=1101, RWED=1111, E=0010
Now reverse the bits:
RE=1010, RED=0010, RWED=0000, E=1101
This gives the 16-bit quantity:
This is the internal representation of the VMS file permission; in
as shown in the sample packet above.
The VMS format probably would also apply to RSX or any other FILES-11
I.2. Handling of Generic Protection
To be used when the two systems are different (and/or do not recognize
or understand each other's local protection codes).
First of all, the book is wrong. This should not be the World
protection, but the Owner protection. The other fields should be set
according to system defaults (e.g. UNIX umask, VMS default protection,
etc), except that no non-Owner field should give more permissions than
the Owner field.
[ [139]Top ] [ [140]Contents ] [ [141]C-Kermit Home ] [ [142]Kermit
Home ]
C-Kermit Program Logic Manual / [143]The Kermit Project /
[144]Columbia University / [145] / 10 April 2004