GnuPG 1.4.5 v, VMS.

Steven M. Schweda sms at
Wed Aug 16 23:30:18 CEST 2006


   I recently did a port of GnuPG 1.4.5 to VMS.  As a result I have some
questions, complaints, and suggestions which may be of some interest. 
For example:

   1. I wrote a VMS-specific entropy gatherer (different from the one HP
supplied in its GnuPG 1.2.3 kit), but I have no confidence in its value. 
Is there any handy test program to which I could attach it for testing
(quality evaluation)?

   2. On VMS (and Tru64), the DEC/Compaq/HP C compiler(s) would normally
emit oodles of PTRMISMATCH and/or PTRMISMATCH1 warnings.  The current
scheme is to disable these warnings, but they could be prevented by some
more careful variable typing and/or some well placed type casts.

   3. Similarly, an oodle of QUESTCOMPARE and QUESTCOMPARE1
informationals could be prevented with some simple (and harmless) type

   4. The VMS compiler is happier with external names shorter than 32
characters.  The VMS-specific "config.h" can "#define" its way around
them, but things would look nicer if "libintl_nl_default_default_domain"
(33) and "libintl_nl_current_default_domain" (33) were a little shorter.

   5. Currently, the "g10defs.h" file is created by the "configure"
script, which is not helpful on VMS (where that script is not used).  It
would be more convenient if the kit included a template file instead,
which could then be edited, either by the "configure" script or by some
VMS-specific procedure, to create the ultimate "g10defs.h" file.

   6.  As VMS currently runs on multiple system types, my normal builder
scheme creates binaries in architecture-specific subdirectories with
names like "alpha", "ia64", and "vax".  On Alpha systems, this can cross
paths with the "mpi/alpha" directory which contains assembly language
source files instead of VMS Alpha binaries.  For the time being, I've
renamed the original "mpi/alpha" to "mpi/src-alpha" in my stuff, and if
some change like that could be main-streamed, I'd be happier.

   As might be expected, there are several changes required to the
source code to allow building and proper operation on VMS.  The changes
I've made could use more testing than I've done, but it might be nice,
eventually, to get some VMS-specific code integrated into the mainstream

   A summary of the changes is included in the release note there: 

   If anyone has any interest, questions, complaints, suggestions, or
whatever, feel free to let me know.


   Steven M. Schweda               sms at antinode-org
   382 South Warwick Street        (+1) 651-699-9818
   Saint Paul  MN  55105-2547

More information about the Gnupg-devel mailing list