[Info-vax] Looking into C-include files on VMS

[Joerg Schilling]

In article <YxCGm.114081$la3.97747 at attbi_s22>,
John E. Malmberg <wb8tyw at qsl.network> wrote:

>> Star offers roughly twice as many features than GNU tar and it is easier to 
>> port than GNU tar too. The only currently still unsupported by star platform I am 
>> aware of is VMS. There are still bugs in GNU tar that I reported in 1993...
>
>There was a lot of work needed to port gnu tar 1.19 to VMS.  For some 
>reason, instead of calling the normal opendir() they are calling an 
>alternate implementation of library calls that was not implemented on VMS.

This may be code that could be reused. Star depends on libfind and libfind
is also used by mkisofs ;-)

>> and it did fail to extract e.g. autoconf/xconfig.h.in without giving even a
>> warning message at the end of the tar run.
>
>The gnv$gnu:[lib]gnv_setup.com is missing the commands
>
>$set process/parse=extended
>
>$define DECC$FILENAME_UNIX_REPORT ENABLE
>$define DECC$FILENAME_UNIX_NOVERSION
>$define DECC$EFS_CASE_PRESERVE ENABLE
>$define DECC$EFS_CHARSET ENABLE
>$define DECC$READDIR_DROPDOTNOTYPE ENABLE

I did add this to my login.com but I see no different behavior.
If I call echo > a.b.c I get an BASH.EXE.1: a.b.c: i/o error


>I have use the gnu tar to extract many tarballs that the vmstar in GNV 
>will not properly read.  That vmstar is also not current.

OK, if there is a way to make a.b.c work, this should help with configure

>Typically configure scripts require:
>
># export GNV_DISABLE_DCL_FALLBACK=1
>
>This is because of a bug in how Bash decides to try a DCL command 
>instead of a UNIX commmand.
>
>And to properly compile main routines that do not explicitly call exit()
>
># export GNV_CC_MAIN_POSIX_EXIT=1

Even with this all set, my configure does not behave correctly.

The test for a Cygwin environment e.g. succeeds even though it should fail
as this code:

#line 1021 "configure" 
#include "confdefs.h" 
 
int main() { 
 
#ifndef __CYGWIN__ 
#define __CYGWIN__ __CYGWIN32__ 
#endif 
return __CYGWIN__; 
; return 0; } 

should abort compilation with am undefined __CYGWIN__ message

I believe that the problem is that "cc" always exits with 0
even if there are compile errors and many of the configure tests
do only check the cc exit code.


>Also dependency-tracking and libtool-lock need to be disabled.

I do not use libtool as libtool did never work correctly for me.


>> I need to work on using "sh" instead of /bin/sh and I ned to find a way
>> to call commands via bash(1) instead of system(3).
>
>/bin/sh should work.  I just tested it on encompasserve.

You are correct, I reported from what I had in mind from another machine that
does not have "/bin"


-- 
EMail:joerg at schily.isdn.cs.tu-berlin.de (home) Jörg Schilling D-13353 Berlin
      js at cs.tu-berlin.de		(uni)  
      joerg.schilling at fokus.fraunhofer.de (work) Blog: http://schily.blogspot.com/
URL:  http://cdrecord.berlios.de/private/ ftp://ftp.berlios.de/pub/schily