Activity for GnuCOBOL

Klaus Siebke posted a comment on discussion Contributions

Klaus Siebke — Sun, 26 Apr 2026 09:18:49 -0000

Hello, I have created a GitHub project demonstrating a proof of concept for a master data maintenance application offering CRUD (create/read/update&delete) functionality and multi-user support developed with GnuCOBOL, VBISAM and GTK3. Two screenshots: And the about screen: Here's the link to the GitHub project: vendorgui3 Have fun with GnuCOBOL Klaus

Ralph Linkletter posted a comment on discussion Help getting started

Ralph Linkletter — Sat, 25 Apr 2026 17:59:52 -0000

The diagnostic is issued with -Wadditional C:\GEDIT\COBOL>cobc -x screen01.cob -Wadditional screen01.cob:24: warning: 'nmbr1' has numeric VALUE without PIC; PIC will be imp lied [-Wadditional] 22 | 02 pic x(01) value space 23 | background-color white lowlight. 24 > 02 nmbr1 pic 9(01) value 1 25 | *> 02 nmbr1 pic x(01) value '1' 26 | background-color white lowlight

Simon Sobisch posted a comment on discussion Help getting started

Simon Sobisch — Sat, 25 Apr 2026 16:39:40 -0000

Nice to know - have you compiled with -Wadditional as well? That warning is not enabled by default.

Ralph Linkletter posted a comment on discussion Help getting started

Ralph Linkletter — Sat, 25 Apr 2026 16:36:10 -0000

When I compiled it with 3.3-dev all is good 0 warnings in compilation group 0 errors in compilation group

Simon Sobisch posted a comment on discussion Help getting started

Simon Sobisch — Sat, 25 Apr 2026 16:28:11 -0000

Dosbox emulates the beep (interrupt) with a speaker signal, GnuCOBOL will beep if the system provides that functionality. Do you build GnuCOBOL on your own? If not: where do you get it from? For esc and function keys -> see runtime.cfg - you can just add a line without a leading # at the place where you want to have a non-default setting - see ESC and function. You also can try to configure the beep to use a different approach and see which is supported by your environment.

Guido Rebry posted a comment on discussion Help getting started

Guido Rebry — Sat, 25 Apr 2026 14:50:58 -0000

When compiling screen01.cob using GnuCobol 3.2 (64 bit, Arnold Trembley package) I get this "false" warning : Q:\COBOL\SOURCE\screen01.COB:24: warning: 'nmbr1' has numeric VALUE without PIC; PIC will be implied [-Wadditional] 22 | 02 pic x(01) value space 23 | background-color white lowlight. 24 > 02 nmbr1 pic 9(01) value 1 25 | *> 02 nmbr1 pic x(01) value '1' 26 | background-color white lowlight When I replace line 24 by line 25 the program compiles without warnings. I know, it's only a warning,...

Steve MacKinney modified a comment on discussion Help getting started

Steve MacKinney — Sat, 25 Apr 2026 13:08:32 -0000

Thanks for the follow up. I installed dosbox and my program runs well on it, including the BEEP function which I use a lot. When entering the basketball play-by-play during a game, the stat guy needs to hear immediately if he enters something invalid. Gnucobol was not going to support BEEP as far as I could tell. I just read thru the previous posts and found the BEEP:BELL note so that may be handled now. It was also hard to identify usage of various function keys - I think ESC was not usable and...

Steve MacKinney posted a comment on discussion Help getting started

Steve MacKinney — Sat, 25 Apr 2026 13:03:10 -0000

Thanks for the follow up. I installed dosbox and my program runs well on it, including the BEEP function which I use a lot. When entering the basketball play-by-play during a game, the stat guy needs to hear immediately if he enters something invalid. Gnucobol was not going to support BEEP as far as I could tell. It was also hard to identify usage of various function keys - I think ESC was not usable and I also use that a lot. If that is not true, I would want to continue the migration project since...

Simon Sobisch posted a comment on discussion Help getting started

Simon Sobisch — Sat, 25 Apr 2026 11:24:19 -0000

So @stevemackinney - how is your migration project going? Did you stumble over anything where a better support from cobc would be helpful?

Simon Sobisch posted a comment on discussion Help getting started

Simon Sobisch — Sat, 25 Apr 2026 10:32:09 -0000

If building on a system with recent GCC you'll likely have to adjust CC/CFLAGS - see [bugs:#1214]. That is not different in GC32. If you have that covered and still get errors (and not build with another isam backend than BDB - see the NEWS comment about ISAM edge cases with other backends), then posting the resulting testsuite.log would be useful. And when you checkout a new version from gc3x branch, then yes, you'll currently get 3.3-dev which is much newer with more features and fixes than gc...

Simon Sobisch posted a comment on ticket #1214

Simon Sobisch — Sat, 25 Apr 2026 10:29:05 -0000

As noted we don't want to enforce C99 and some compilers even don't have that option - we'd want to check if the compiler setup as-is does compile the relevant code, and only if not - and no std is given - try to add it: Adding it to default CFLAGS will need a test in configure in any case - so we should first check if the void function declaration and pointer use works or not - and only try to work around that by specifying a std as-needed (and then try gnu17 / c17 / gnu11 /c11 / gnu99 / gnu89 /...

Rich Di Iulio posted a comment on discussion Help getting started

Rich Di Iulio — Fri, 24 Apr 2026 20:35:29 -0000

Thank you Simon When I did option #1. I got a mismatch of software. Looking at HACKING. I saw a option ./autogen.sh install. I invoked the command and i was able to run configure and make successfully. Ran make check and make test. There were a number of errors. I installed the compiler onto my Manjaro VM. When I checked the version it was gnucobol-3.3.-dev.0 I thought there was an update to gnucobol-3.2. I expected that version when I check it out. Am I missing something? Should I attach the log...

Simon Sobisch posted a comment on discussion Help getting started

Simon Sobisch — Fri, 24 Apr 2026 07:58:01 -0000

configure and other generated files aren't part of the repository since... GC 1.1 or so. You have two options: 1) build from version control -> after checkout/clone/download execute autogen.sh (creates the additional files, needs more tools - see HACKING for the list); this works with any version (also older ones or "tags") 2) get a nightly tarball (or even binary), the places to look for are https://sourceforge.net/projects/gnucobol/files/gnucobol/nightly_snapshots/ and nowadays also the OCamlPro...

Rich Di Iulio posted a comment on discussion Help getting started

Rich Di Iulio — Fri, 24 Apr 2026 00:11:12 -0000

Hi In the past I was able to use svn to checkout a version of GnuCOBOL. Here is the command line: svn checkout https://svn.code.sf.net/p/gnucobol/code/branches/gnucobol-3.x gnucobol-3.2 It downloads files, however; there is no configure file. Has something changed? I am looking to download the latest gnucobol-3.2. I also would like to download the latest of gnucobol-3.3-dev. I am working on Manjaro. I have been able to compile successfully vbisam2.2 and visam-2.2. I appreciate any help. Regards,...

Simon Sobisch posted a comment on discussion The Lounge

Simon Sobisch — Thu, 23 Apr 2026 19:08:05 -0000

Of course there are hallucinations and missing information in that LLM generated text... there is no libcob.conf but runtime.cfg there is no COB_EBCDIC_OVERRIDE runtime parameter, instead there is the CODE-SET clause in plain COBOL that can be used to do on-the-fly conversion the file part and the general post miss to note the different collating sequence - which again are both possible to explicit code in COBOL (and with GnuCOBOL 3.x+ also possible to pass on the command line / dialect configuration)...

Mickey White posted a comment on discussion The Lounge

Mickey White — Thu, 23 Apr 2026 18:07:39 -0000

I actually like Claude. I have a GAB subscription and that is good enough for me. Ralph, just a note, when I start a session about a cobol project or issue, I give AI my expectaions and specifications on coding. I usually just paste this Text or MarkDown file into the AI and then proced with my coversation... Here attached is the file I use, and it is a continuing process to add or subtract as I use AI. Note, I am trying to learn C so not much cobol going on these few months... Attahced:

Mickey White posted a comment on discussion The Lounge

Mickey White — Thu, 23 Apr 2026 17:53:41 -0000

I posted the one here, but the longer one would not let me post on linked in. I will include it here for the peoples ! ~~ Response to the WINCOBOL/zOS Migration Post This is a well-written and technically accurate critique. Let me engage with it honestly rather than defensively. What the Author Gets Right The EBCDIC problem is genuinely hard. The commingling of: EBCDIC character fields Packed decimal (COMP-3) Big-endian binary (COMP, COMP-4, COMP-5) Arrays (OCCURS) REDEFINES clauses ...in a single...

Ralph Linkletter posted a comment on discussion The Lounge

Ralph Linkletter — Thu, 23 Apr 2026 17:27:26 -0000

Hey Mickey can you post this as a comment to a post I made on LinkedIn? Url below: https://www.linkedin.com/posts/ralph-linkletter-0a1a6218b_ai-driven-mainframe-migrations-a-truthful-activity-7452777690849439744-wrlP?utm_source=share&utm_medium=member_desktop&rcm=ACoAACzUG_kBgHDzwjeXOhYmzGBLqRK7rRNH8Bo Also consider the longer Claude analysis that you emailed to me. It seems inappropriate for me to post these as comment to my own post - agreed ?

Mickey White posted a comment on discussion The Lounge

Mickey White — Thu, 23 Apr 2026 16:22:25 -0000

Claude agrees: You’re welcome. The post makes legitimate technical points that deserve honest acknowledgment rather than defensive deflection. The author clearly has deep hands-on experience with the actual mechanics of zOS - not just theoretical knowledge. The packed decimal / big-endian binary / REDEFINES / VSAM combination is genuinely a multi-layered problem that no AI tool is going to auto-solve at production scale. Anyone selling that story to a CIO who has never seen a hex dump of a VSAM record...

Ralph Linkletter posted a comment on discussion The Lounge

Ralph Linkletter — Thu, 23 Apr 2026 15:27:53 -0000

April 21, 2026 "Claude AI - COBOL" and rescuing the IBM mainframe. What nonsense! What is the source of this "magical thinking / wishful thinking”? "Mainframe”, in this document is defined as IBM zOS. The Linux / Unix operating systems running under zOS are not considered to be "Mainframe". Take a peek at why this Claude rescue is nonsensical. Linux and Windows are the target operating systems of mainframe migrations. This is a tiny slice of an actual mainframe VSAM data file. See attachment As an...

Simon Sobisch posted a comment on discussion Help getting started

Simon Sobisch — Wed, 22 Apr 2026 07:39:37 -0000

It all depends what license concerns you have (just my understanding, consult your lawyer if the business depends on it): if you use GnuCOBOL only for your in-house software - there's no problem with using AGPL ever, no matter how you use it (it is your code and you are your only customer for that software) if you use it to provide a web frontend to customers then you'd have to make the code ( everything using that) available via that frontend as well - no problem for BDB+GnuCOBOL+dependencies, but...

Anonymous posted a comment on discussion Help getting started

Anonymous — Wed, 22 Apr 2026 07:12:06 -0000

What if you build GnuCOBOL with BDB v5.3.28? Wouldn't that eliminate licensing concerns because it was the last release before the license was changed to AGPLv3? Anon

Mickey White posted a comment on discussion The Lounge

Mickey White — Mon, 20 Apr 2026 15:38:33 -0000

https://www.youtube.com/watch?v=FPR-aPZAhZ4

Vincent (Bryan) Coen committed [r5673] on Code

Vincent (Bryan) Coen — Fri, 17 Apr 2026 12:42:17 -0000

1222

Vincent (Bryan) Coen modified ticket #1222

Vincent (Bryan) Coen — Fri, 17 Apr 2026 12:41:38 -0000

The *CBL Directive is missing

Vincent (Bryan) Coen posted a comment on ticket #1222

Vincent (Bryan) Coen — Fri, 17 Apr 2026 12:41:24 -0000

I have corrected errors: Last param for the must be None Comment added for CBL saying that is it NOT implemented at least for v3.2. I am not so sure that it should be specified if it is not in use and the LIST and MAP is unlikely to ever be. So whats the point of it.

Simon Sobisch posted a comment on ticket #470

Simon Sobisch — Fri, 17 Apr 2026 12:15:28 -0000

It seems this ticket has been "lost" on the way. We found out that there's no "very easy" solution as we have to consider both the DEBUG-ITEM stuff and same field names but a different reference (occurs). Do you want to come back to this or should we end our findings with "we'd generate one seldom (COBOL-code), but unnecessary memmove less - but need to add a bunch of code for doing so" and close this ticket? Even if we do so... do we have useful tests that we can add to cover the current code better...

Simon Sobisch posted a comment on ticket #470

Simon Sobisch — Fri, 17 Apr 2026 12:09:47 -0000

@dhugonnard this patch is unrelated to this ticket - if we don't have it in another place please post it whereever it is appropriate

Eugenio Di Lorenzo posted a comment on ticket #1222

Eugenio Di Lorenzo — Fri, 17 Apr 2026 11:47:44 -0000

I had already tested @OPTIONS, which exists for both Fujitsu and AcuCOBOL (with different parameters !). The clause is detected and accepted, but the result is always: warning: ignoring unknown directive: '@options FREE APOST' [-Wothers] The clause *CBL or *CONTROL had been documented in Ed Hart's GRAMMAR.

Simon Sobisch posted a comment on ticket #1222

Simon Sobisch — Fri, 17 Apr 2026 11:18:05 -0000

Just a note: @OPTIONS directive from Fujitsu is also "kind of supported" (I think it currently is just ignored, but adding support for the ones we have CBL or $ or >> support for is "only some tedious work of mostly copy+paste" ) - you hopefully find something in our testsuite for that. ... speaking of where to get data: how was the CDF-CBL list info resolved? Does anyone know if the MF dollar list is complete? Checking for possible missing items likely means inspecting ppparse.y (and pplex.l).....

Eugenio Di Lorenzo modified ticket #1222

Eugenio Di Lorenzo — Fri, 17 Apr 2026 11:14:50 -0000

The *CBL Directive is missing

Eugenio Di Lorenzo created ticket #1222

Eugenio Di Lorenzo — Fri, 17 Apr 2026 11:09:46 -0000

The *CBL Directive is missing

Eugenio Di Lorenzo posted a comment on discussion GnuCOBOL

Eugenio Di Lorenzo — Fri, 17 Apr 2026 07:49:56 -0000

Tildes also don't exist on the Italian keyboard. (And neither do curly brackets). To create tildes, however, in SourceForge we have the button. If you mark a block of multiple lines, then using inserts three tildes at the beginning of the block and three tildes at the end of the block. If, however, you mark a block on a single line or even a single word and use , a backtick is inserted at the beginning of the word and one at the end of the word.

Denis HUGONNARD-ROCHE posted a comment on ticket #72

Denis HUGONNARD-ROCHE — Fri, 17 Apr 2026 07:22:07 -0000

Ok . By the way i manage to have a version of cob_mpf_exp that on my pc is between 4 and 5 times faster .

Simon Sobisch posted a comment on ticket #72

Simon Sobisch — Thu, 16 Apr 2026 21:44:54 -0000

Feel free to do so (note: having at least a "hint" here is always good, as GitHub will only be a mirror - but learning git + PR workflows is good, as that will be nearly identical when we migrate from here to codeberg late[r] this year). I hope to get back to the PRs (and other open issues) when some of the load at work is down again.

Simon Sobisch posted a comment on discussion GnuCOBOL

Simon Sobisch — Thu, 16 Apr 2026 17:34:05 -0000

Thanks for letting me know - obviously default keyboard layouts on Windows for Italy don't have that character - but not even the tilde? Both are parts of ASCII 7 bit, so that comes as a surprise. Backtick is classically used on any *nix as that creates a subshell and the tilde is the home directory, so also used often. Seems like https://github.com/ilpianista/itlinux-win-keyboard fixes both with the expected AltGr shortcuts.

Denis HUGONNARD-ROCHE posted a comment on ticket #72

Denis HUGONNARD-ROCHE — Thu, 16 Apr 2026 14:34:02 -0000

Hi Simon do you want me to integrate the patch for ln an for the ticjer #1206 in my github repo in order to create a pull request (now that i begin to use git & github serioulsy , may ba it would be good to have all in github) Denis

Eugenio Di Lorenzo posted a comment on discussion GnuCOBOL

Eugenio Di Lorenzo — Thu, 16 Apr 2026 13:53:23 -0000

Okay, I get it, but this is really weird. I've never used the backstick character before ! I saw that you can only get it by typing Alt + 96.

Simon Sobisch posted a comment on discussion GnuCOBOL

Simon Sobisch — Thu, 16 Apr 2026 12:48:06 -0000

three (or more) backticks `, but yes, it seems to work on the tildes ~ as well (it used to need a newline + : + name, something you may recognize in old posts)

Eugenio Di Lorenzo posted a comment on discussion GnuCOBOL

Eugenio Di Lorenzo — Thu, 16 Apr 2026 12:29:46 -0000

I don't understand what characters you used before "cobolfree". Anyway, I used three tildes (using ther button ) and added the string "cobolfree" after first marker and it seems to work just the same.

Eugenio Di Lorenzo modified a comment on discussion GnuCOBOL

Eugenio Di Lorenzo — Thu, 16 Apr 2026 12:14:40 -0000

ADD ... TO .... GIVING: a. In the ISO standard, the "TO" clause is mandatory. b. In IBM's COBOL documentation, the "TO" clause is indicated as optional. c. GnuCOBOL also allows the "TO" clause to be optional, and this is documented in the PG. see attached diagrams. The example program demonstrates that GnuCOBOL compiles without errors. It's unclear what the meaning of it being optional might be. >>SOURCE FREE IDENTIFICATION DIVISION. PROGRAM-ID. TESTADD2. DATA DIVISION. WORKING-STORAGE SECTION. 01...

Simon Sobisch posted a comment on discussion GnuCOBOL

Simon Sobisch — Thu, 16 Apr 2026 10:12:14 -0000

side note: if you surround your code by the correct matching syntax marker, you get better highlighting - please edit your post to test that Instead of ~~~ ADD AAA 10 TO 22 GIVING GGG DDD *> comment here ADD AAA 10 22 GIVING GGG DDD ~~~ which renders as ADD AAA 10 TO 22 GIVING GGG DDD *> comment here ADD AAA 10 22 GIVING GGG DDD try ```cobolfree ADD AAA 10 TO 22 GIVING GGG DDD *> comment here ADD AAA 10 22 GIVING GGG DDD ``` which renders as ADD AAA 10 TO 22 GIVING GGG DDD *> comment here ADD AAA...

Simon Sobisch posted a comment on discussion GnuCOBOL

Simon Sobisch — Thu, 16 Apr 2026 10:09:35 -0000

It is optional with ISO as well (the TO is not underlined), so we luckily don't have to add a dialect configuration :) ADD ... GIVING adds a list of elements (which may be only 1 element) to another element, stores the result in an intermediate result and then stores it into a list of other elements. It does make no difference to add a list of "one more" element to logical zero to - the result in the intermediate is still the same. ADD AAA 10 TO 22 GIVING GGG DDD ADD AAA 10 22 GIVING GGG DDD

Mickey White posted a comment on discussion Contributions

Mickey White — Wed, 15 Apr 2026 23:17:08 -0000

Perhaps more details would be needed to know what the issue is?

Vincent (Bryan) Coen committed [r5672] on Code

Vincent (Bryan) Coen — Wed, 15 Apr 2026 18:08:13 -0000

1221

Vincent (Bryan) Coen modified ticket #1221

Vincent (Bryan) Coen — Wed, 15 Apr 2026 18:07:10 -0000

Some fixes on Chapter 2 and SUBTRACT

Eugenio Di Lorenzo modified a comment on discussion GnuCOBOL

Eugenio Di Lorenzo — Wed, 15 Apr 2026 17:23:53 -0000

Eugenio Di Lorenzo posted a comment on discussion GnuCOBOL

Eugenio Di Lorenzo — Wed, 15 Apr 2026 17:14:34 -0000

Eugenio Di Lorenzo created ticket #1221

Eugenio Di Lorenzo — Wed, 15 Apr 2026 17:00:58 -0000

Some fixes on Chapter 2 and SUBTRACT

Xavier Del Campo Romero modified a comment on ticket #1214

Xavier Del Campo Romero — Wed, 15 Apr 2026 14:23:48 -0000

@sf-mensch what do you think about the following patch?

Xavier Del Campo Romero modified a comment on ticket #1214

Xavier Del Campo Romero — Wed, 15 Apr 2026 14:22:47 -0000

@sf-mensch what do you think about the following patch? ``` Index: configure.ac =================================================================== --- configure.ac (revision 5668) +++ configure.ac (working copy) @@ -714,6 +714,21 @@ AC_STRUCT_TM AC_CHECK_TYPES([sig_atomic_t], [], [], [[#include ]]) +curr_cflags="$curr_cflags" +CFLAGS="$CFLAGS -std=gnu99" +AC_MSG_CHECKING([for -std=gnu99 support]) +AC_COMPILE_IFELSE([AC_LANG_SOURCE([[ if !defined(STDC) || !defined(STDC_VERSION)...

Xavier Del Campo Romero posted a comment on ticket #1214

Xavier Del Campo Romero — Wed, 15 Apr 2026 14:22:32 -0000

@sf-mensch what do you think about the following patch? ```patch Index: configure.ac =================================================================== --- configure.ac (revision 5668) +++ configure.ac (working copy) @@ -714,6 +714,21 @@ AC_STRUCT_TM AC_CHECK_TYPES([sig_atomic_t], [], [], [[#include ]]) +curr_cflags="$curr_cflags" +CFLAGS="$CFLAGS -std=gnu99" +AC_MSG_CHECKING([for -std=gnu99 support]) +AC_COMPILE_IFELSE([AC_LANG_SOURCE([[ if !defined(STDC) || !defined(STDC_VERSION)...

Vincent (Bryan) Coen posted a comment on discussion The Lounge

Vincent (Bryan) Coen — Sun, 12 Apr 2026 16:07:48 -0000

This is all documented but little is always for standard Micros (there are exceptions for some with 64 and 128 bit CPU's). IT is hardware dependent. COMPUTATIONAL or COMP as against COMP-1 thru 15 and X etc. RTFM's. GC is NOT designed for use as standard with M/F's so mileage may vary. I did try and install it on a OS390 system ssome years back but underlining packages was not available to support it but that may now have changed but in any event use FREE format could well be an issue with PDSE etc....

Ralph Linkletter modified a comment on discussion The Lounge

Ralph Linkletter — Sun, 12 Apr 2026 15:19:25 -0000

"BINARY = COMP " Which COMP ? big or little endian ? "BINARY-CHAR, BINARY-SHORT, BINARY-LONG . . . Where did this syntax originate? All the above are little endian How confusing can it be made ? BINARY with a PICTURE CLAUSE is big endian BINARY-xxxxx without a PICTURE CLAUSE is little endian COMP-5 on zOS is big endian that infers implicitly NOTRUNC The screen shots are from AI ? Is there a "C" data type for big endian ? The screen shots identify the big endian formats as int / uint with scalar -...

Ralph Linkletter posted a comment on discussion The Lounge

Ralph Linkletter — Sun, 12 Apr 2026 15:18:36 -0000

Vincent (Bryan) Coen posted a comment on discussion The Lounge

Vincent (Bryan) Coen — Sun, 12 Apr 2026 13:34:09 -0000

BINARY = COMP defined by quantity of 9's and presence or not of S - see 6.9.52. BINARY-CHAR = 1 byte 0 - 255 depending on usage of SIGNED / UNSIGNED. See 7.1.6.3. On 12/04/2026 03:49, Ralph Linkletter wrote: COMP-5 and COMP/COMP- have the same "C" typing ? COMP / COMP-4 are big endian - is there such a "C" type ? I do not understand BINARY-CHAR (signed or unsigned) I guess it would be PIC 9(02) COMP-X. If so a compile would need -fbinary-size=1-8 PIC X(01) COMP-X If I recall correctly there is zOS...

Ralph Linkletter modified a comment on discussion The Lounge

Ralph Linkletter — Sun, 12 Apr 2026 03:55:34 -0000

COMP-5 and "COMP / COMP- 4 " have the same "C" typing ? COMP / COMP-4 are big endian - is there such a "C" type representing big endian ? I do not understand BINARY-CHAR (signed or unsigned) I guess it would be PIC 9(02) COMP-X. If so a compile would need -fbinary-size=1-8 PIC X(01) COMP-X If I recall correctly there is zOS BINARY as well. This BINARY form has no suffix - "-SHORT for example" The zOS form of binary is big endian. The above BINARY form is not provided for in any table on this thread....

Ralph Linkletter modified a comment on discussion The Lounge

Ralph Linkletter — Sun, 12 Apr 2026 03:55:33 -0000

Ralph Linkletter posted a comment on discussion The Lounge

Ralph Linkletter — Sun, 12 Apr 2026 02:49:08 -0000

COMP-5 and COMP/COMP- have the same "C" typing ? COMP / COMP-4 are big endian - is there such a "C" type ? I do not understand BINARY-CHAR (signed or unsigned) I guess it would be PIC 9(02) COMP-X. If so a compile would need -fbinary-size=1-8 PIC X(01) COMP-X If I recall correctly there is zOS BINARY as well. This BINARY form has no suffix - "-SHORT for example" The zOS form of binary is big endian. The above BINARY form is not provided for in any table on this thread. IBM COBOL is 90+ % of the universe...

Mickey White posted a comment on discussion The Lounge

Mickey White — Sun, 12 Apr 2026 00:42:38 -0000

*not verified

Mickey White posted a comment on discussion The Lounge

Mickey White — Sun, 12 Apr 2026 00:33:11 -0000

here is an AI version:

Vincent (Bryan) Coen committed [r5671] on Code

Vincent (Bryan) Coen — Sat, 11 Apr 2026 18:16:37 -0000

1220

Vincent (Bryan) Coen modified ticket #1220

Vincent (Bryan) Coen — Sat, 11 Apr 2026 18:15:11 -0000

PICTURE U is missing

Eugenio Di Lorenzo modified ticket #1220

Eugenio Di Lorenzo — Sat, 11 Apr 2026 16:51:33 -0000

PICTURE U is missing

Eugenio Di Lorenzo created ticket #1220

Eugenio Di Lorenzo — Sat, 11 Apr 2026 16:43:29 -0000

PICTURE U is missing

Ralph Linkletter posted a comment on ticket #1219

Ralph Linkletter — Sat, 11 Apr 2026 16:17:40 -0000

Why was it an intermittent failure ? Should it not have failed all the time ?

Guido Rebry posted a comment on discussion Help getting started

Guido Rebry — Sat, 11 Apr 2026 14:01:56 -0000

And it works like a charm! Tyvm for the perfect explanation.

Ralph Linkletter modified a comment on discussion The Lounge

Ralph Linkletter — Sat, 11 Apr 2026 02:23:50 -0000

It would be great if someone also added the comp-? to the documents as well as the BINARY stuff. There is a set of arithmetic COBOL definitions missing from this document. I have no idea what a BINARY-CHAR is. PIC X(01) comp-x? The COMP-5 declarations have "C" counter parts with "C" scalars BINARY less any suffix is big endian - how are bytes allocated to BINARY (big endian) PIC9(04), 9(08, 9(15) comp. Indicating the number of bytes allocated to each Discuss -fnotrunc / -ftrunc and its effect on...

Ralph Linkletter posted a comment on discussion The Lounge

Ralph Linkletter — Sat, 11 Apr 2026 02:21:20 -0000

It would be great if someone also added the comp-? to the documents as well as the BINARY stuff. There is a set of arithmetic COBOL definitions missing from this document. I have no idea what a BINARY-CHAR is. PIC X(01) comp-x? The COMP-5 declarations have "C" counter part with "C" scalars BINARY less any suffix is big endian - how are bytes allocated to BINARY (big endian) PIC9(04), 9(08, 9(15) comp. Indicating the number of bytes allocated to each Discuss -fnotrunc / -ftrunc and its effect on computations...

Michael Del Solio posted a comment on discussion Help getting started

Michael Del Solio — Fri, 10 Apr 2026 22:51:10 -0000

Please have a look on my previous posts above (you may have missed it): https://sourceforge.net/p/gnucobol/discussion/help/thread/59e651b2c7/?limit=25#eb3e/0a15/fed0/0161 I figured out, how we can do it with the WinBGm-library with SuperBOL-AIO and provided a little howto with examples and the commands to compile it. Screenshot is also attached there. The second example opens a new window with the graphics output (lines, rectangle and circle) :-)

Eugenio Di Lorenzo posted a comment on discussion Help getting started

Eugenio Di Lorenzo — Fri, 10 Apr 2026 21:15:44 -0000

It seems the issue is getting complicated. I'd like to try to sort out and/or simplify what's emerged so far. A graphics library is required, and the one that works with both pdcurses (used by Arnold's GnuCOBOL) and ncursesw (used by SuperBOL AIO) is the Raylib library. It's unclear whether, as I hope, there's a ready-to-download railib.lib library for WINDOWS 11, or whether it has to be built by compiling from source code. If there's a ready-to-download raylib.lib library to use with GnuCOBOL (Arnold...

László Erdős posted a comment on discussion The Lounge

László Erdős — Fri, 10 Apr 2026 20:25:37 -0000

Yes, it is from the GnuCobol Programmers Guide. I called a C function with double parameter, and without to check the mapping in the documentation, I defined a COBOL variable with BINARY-DOUBLE data type. Analouge like the other mappings. But of course it didn't work. Only the "double" correspondes not to the "double". We need here COMP-2. Why it is so, I don't know? It took some time until I found it out. :)

Eugenio Di Lorenzo posted a comment on discussion The Lounge

Eugenio Di Lorenzo — Fri, 10 Apr 2026 20:14:06 -0000

See also attached image from: https://svn.code.sf.net/p/gnucobol/code/external-doc/GnuCOBOL_C_Interaction.pdf

Vincent (Bryan) Coen posted a comment on discussion The Lounge

Vincent (Bryan) Coen — Fri, 10 Apr 2026 19:31:45 -0000

As found in the GnuCobol Programmers Guide.

Chuck Haatvedt modified ticket #1219

Chuck Haatvedt — Fri, 10 Apr 2026 18:06:19 -0000

COBC signal SIGSEGV

Chuck Haatvedt posted a comment on ticket #1219

Chuck Haatvedt — Fri, 10 Apr 2026 18:05:03 -0000

Here is the patch for this issue. in cobc.c in the cobc_xref_set_receiving function added the following to check for a NULL field name and return if it is NULL. *** F:/GNUCOBOL-33-SVN/cobc/cobc.c Tue Dec 2 14:17:48 2025 --- F:/AA-minGW32-static/cobc-xref-fix.c Fri Apr 10 12:27:42 2026 *************** *** 6128,6129 **** --- 6128,6132 ---- target_fld = CB_FIELD_PTR (target); + if (!target_fld->name) { + return; + } target_fld->count++; Chuck Haatvedt

Vincent (Bryan) Coen posted a comment on discussion GnuCOBOL

Vincent (Bryan) Coen — Fri, 10 Apr 2026 15:01:20 -0000

The simplest way is to use a scheduler such as Linux cron running at the required time and issue a ftp command of your choice or failing that do a rsync to the same site address. Make sure you check the return code of both methods to confirm it worked. I use both methods under Linux but if you are a wndoz user there is its own cron replacement. Note that cron will run even if a system failure occurs of the sending system in that it will run when the system has been started back up. Vince On 10/04/2026...

Ralph Linkletter posted a comment on ticket #1219

Ralph Linkletter — Fri, 10 Apr 2026 14:15:13 -0000

The abend is caused by the -Xref compiler option when the source references an undeclared data element. *** > MOVE 0 TO XENV-CNT*** The attached TEMP.CMD file can be used to reproduce the failure. Recall that the failure is intermittent. But executing the .CMD file 4 or so times will produce the failure. cobc -m ^ -tCEEENV.LST ^ -Xref ^ CEEENV.COB

peter stanyer posted a comment on discussion GnuCOBOL

peter stanyer — Fri, 10 Apr 2026 14:09:18 -0000

Finally found the code hidden away in a copy book. The program calls the 'STRING "curl -s -T' FTP command via system as suggested contained in WS-COMMAND. Its moving over the file and stripping any spaces off the fields by the look of it. Also seems to do this every day when the final user logs out of the system. It seems to check if the random indexed file is open then when a user closes the program it ignores the routine but if the file is not open by anyone it send the sequential file as the final...

Simon Sobisch posted a comment on discussion GnuCOBOL

Simon Sobisch — Fri, 10 Apr 2026 12:54:49 -0000

For automation: moat likely something like a scheduled task (from the OS) was used, otherwise there would be a program sleeping then checking current time and tasks, executing an external call whenever needed.

Simon Sobisch posted a comment on discussion GnuCOBOL

Simon Sobisch — Fri, 10 Apr 2026 12:52:01 -0000

you can use something similar to CALL "SYSTEM" USING "ftp file server" with most likely the using part in a variable.

peter stanyer posted a comment on discussion GnuCOBOL

peter stanyer — Fri, 10 Apr 2026 10:45:56 -0000

I am updating some very old 1984 rm/cobol stock control software which has some missing source code. One part missing is the software that saves a line sequential file called 'mitsulost.csv' to a remote server. This is not a shared local server or mapped drive, it is a remote sql server and the file is simply stored on the remote drive presumably similar to using some sort of ftp. The file uploaded contains lost part numbers and is approximately 75-100 lines long. The remote system reads these files...

Simon Sobisch modified ticket #506

Simon Sobisch — Thu, 09 Apr 2026 19:53:11 -0000

patch for GCSORT (improve and check COBOL performance)

Ralph Linkletter posted a comment on ticket #1219

Ralph Linkletter — Thu, 09 Apr 2026 18:23:02 -0000

The command line: c:\IBZANIM\COBOL>C:\IBZANIM\WINGNU\BIN\cobc \IBZANIM\COBOL\CEEENV.COB -tC:\IBZANIM\LISTINGS\CEEENV.L ST -oC:\IBZANIM\LOADLIB\CEEENV -fgoto-entry=ok -fno-ec=all -LC:\WINGNU\INCLUDE -ffold-call=upper -L C:\IBZANIM\WINGNU\INCLUDE -O0 -fassign-clause=ibm -fno-gen-c-decl-static-call -fbinary-size=2-4-8 -- tsymbols -fsign=ebcdic -fodoslide -fimplicit-init -fperform-osvs -fmove-ibm -fhostsign -fnotrunc -fn o-recursive-check -fdefaultbyte=00 -fsticky-linkage -fsource-location -Xref -A "-include...

Ralph Linkletter created ticket #1219

Ralph Linkletter — Thu, 09 Apr 2026 18:20:33 -0000

COBC signal SIGSEGV

Sauro Menna posted a comment on ticket #506

Sauro Menna — Thu, 09 Apr 2026 17:21:19 -0000

Hi Simon, I've updated the GCSORT version by incorporating the files you sent me. I've released the updates in GCSORT version 1.04.08 (2025-09-21). Thanks for your suggestions and advice. Sauro

Saurabh Kumar posted a comment on discussion Contributions

Saurabh Kumar — Thu, 09 Apr 2026 08:40:28 -0000

Hello team! It looks like the CI might be broken for macOS, as the failures reported there are not reproducible locally with the same commands. I noticed there are some open PRs that fix it. Could someone please take a look and consider reviewing and merging them to help unblock progress? Thanks!

László Erdős posted a comment on discussion Help getting started

László Erdős — Wed, 08 Apr 2026 16:42:24 -0000

Hello, The Raylib C library is also based on BGI. https://en.wikipedia.org/wiki/Raylib https://github.com/raysan5/raylib There are Raylib bindings and wrappers for many programming languages: https://github.com/raysan5/raylib/blob/master/BINDINGS.md Also for COBOL: https://codeberg.org/glowiak/raylib-cobol Our GNU Cobol Contributions also include an example using Raylib (author: Giancarlo Canini): https://sourceforge.net/p/gnucobol/contrib/HEAD/tree/trunk/samples/raylib_painter However, the main...

Michael Del Solio posted a comment on discussion Help getting started

Michael Del Solio — Wed, 08 Apr 2026 13:23:27 -0000

Now, we have some more functions available: >>SOURCE FREE IDENTIFICATION DIVISION. PROGRAM-ID. GFXDEMO. DATA DIVISION. WORKING-STORAGE SECTION. 78 C-BLACK VALUE 0. 78 C-BLUE VALUE 1. 78 C-GREEN VALUE 2. 78 C-CYAN VALUE 3. 78 C-RED VALUE 4. 78 C-MAGENTA VALUE 5. 78 C-BROWN VALUE 6. 78 C-LIGHTGRAY VALUE 7. 78 C-DARKGRAY VALUE 8. 78 C-LIGHTBLUE VALUE 9. 78 C-LIGHTGREEN VALUE 10. 78 C-LIGHTCYAN VALUE 11. 78 C-LIGHTRED VALUE 12. 78 C-LIGHTMAGENTA VALUE 13. 78 C-YELLOW VALUE 14. 78 C-WHITE VALUE 15. 01...

Michael Del Solio modified a comment on discussion Help getting started

Michael Del Solio — Wed, 08 Apr 2026 08:34:54 -0000

Ok, I think we are on the good way with winbgm/putpixel() on SuperBOL-AIO-distribution :-) 1) Library (always check it with virustotal.com) and graphics.h from: https://github.com/dbautsch/winbgim/releases/tag/6.0a 2) Rename libbgi64.a to libbgi.a and put it into: C:\Users\YOURUSER\AppData\Local\GnuCOBOL\mingw64\lib\gcc\x86_64-w64-mingw32\13.2.0 3) Put graphics.h (from the Source Code ZIP file) into: C:\Users\YOURUSER\AppData\Local\GnuCOBOL\mingw64\lib\gcc\x86_64-w64-mingw32\13.2.0\include 4) Compile...

Michael Del Solio posted a comment on discussion Help getting started

Michael Del Solio — Wed, 08 Apr 2026 08:30:02 -0000

Vincent (Bryan) Coen committed [r5670] on Code

Vincent (Bryan) Coen — Wed, 08 Apr 2026 00:04:42 -0000

1218

Vincent (Bryan) Coen modified ticket #1218

Vincent (Bryan) Coen — Wed, 08 Apr 2026 00:02:53 -0000

Fixes to chapter 6 and 8

Vincent (Bryan) Coen posted a comment on ticket #1218

Vincent (Bryan) Coen — Wed, 08 Apr 2026 00:02:53 -0000

done

Chuck Haatvedt posted a comment on discussion Help getting started

Chuck Haatvedt — Tue, 07 Apr 2026 23:14:27 -0000

PDCursesMOD does provide a Windows graphic version called WINGUI. With this you should be able to access individual pixels using the setpixel function. There are some serious considerations How to Access Pixels in WinGUI To draw or manipulate individual pixels, you must bypass the curses abstraction and use the underlying Windows API: Get the Device Context (HDC): You need a handle to the window's device context to perform graphical operations. In the WinGUI port, you can often obtain the relevant...

Michael Del Solio posted a comment on discussion Help getting started

Michael Del Solio — Tue, 07 Apr 2026 22:48:57 -0000

Thank you very much. I searched now for a 64-bit version and found this one for MinGW: https://github.com/dbautsch/winbgim/releases/tag/6.0a The binary libbgi64.a looks ok while checked with virustotal (please recheck if you want to use it). We need to extract the graphics.h from the file "Source code (zip)" and put it into include folder of the SuperBOL-installation ... and we need to put ibbgi64.a into the libs folder... Tomorrow (or within the next days) I will try to compile the example from...

Eugenio Di Lorenzo created ticket #1218

Eugenio Di Lorenzo — Tue, 07 Apr 2026 22:09:24 -0000

Fixes to chapter 6 and 8

Vincent (Bryan) Coen committed [r5669] on Code

Vincent (Bryan) Coen — Tue, 07 Apr 2026 20:30:44 -0000

1215 & 1216