26-Nov-2005

Seti code and what's involved
Or: How "Open" (as in "Open source") is abused.
BOINC (Berkeley Open Infrastructure for Network Computing) can be downloaded, as well as the SETI client. But not directly, at least, no longer. Today, the whole code can only be downloaded using CVS - another Unix "standard" (therefore nicknamed " Open source" ) which I of course don't have on the system. Remember, this is OpenVMS, not Unix. However, it does exist, even on Freeware CD 4 - but a newer version is available, so I downloaded the whole stable version (1.11.21). There is indeed VMS support; that is: there is a directory named VMS, and there are files present - commandprocedures and MMS files - to build the product.
Alas, it doesn't work that way.
First of all, one executable : minigzip.exe, is causing trouble, it seems to require an X11 library that does not exist. But it was found that in the commandprocedure that builds straight forward using C, this source is commented out for buiilding, but the DECRIP.MMS file still contains the reference. And DECRIP.MMS is used, since I have MMS installed.
Easy solution: comment these lines out.

Next problem is compiling diff module, since an include file seems to be missing: fnmatch.h. DIFF.C contains this code:

#ifdef HAVE_FNMATCH
# include /* This is supposed to be available on Posix systems */
#else /* HAVE_FNMATCH */
# include "fnmatch.h" /* Our substitute */
#endif /* HAVE_FNMATCH */

and since the list showed "Our substitute", the conclusion is that HAVE_FNMATCH was not defined.
Looking for this file on the exepected location (SYS$LIBRARY) showed it does not exist there, not in any of the text libraries. But I could find it in the GNV environment, one in the [.DIFF] directory. So I copied that to [.DIFF], and now compilation went on but gave warnings:

int link (const char *__existing, const char *__new);
...........^%CC-W-MISMATPARAM, In this declaration, parameter 1 has a different type than specified in an earlier declaration of this function.
at line number 663 in module UNISTD of text library SYS$COMMON:[SYSLIB]DECC$RTLDEF.TLB;2

Nasty, but I choose to leave that for a moment; it's generic, and shows up in a lot of places.

But finally, it ended with:

&& (VMS_filename_classes[(unsigned char) *n1]
...............^
%CC-E-UNDECLARED, In this statement, "VMS_filename_classes" is not declared.
at line number 669 in file DKB600:[cvs_base.CVS-1_11_21.VMS]FILESUBR.C;1

last_component (path)
^
%CC-W-NOTCOMPAT, In this declaration, the type of "last_component" is not compatible with the type of a previous declaration of "last_component" at line number500 in file DKB600:[cvs_base.CVS-1_11_21.SRC]CVS.H;1.
at line number 906 in file DKB600:[cvs_base.CVS-1_11_21.VMS]FILESUBR.C;1

Because of the error, the commandfile stopped....
This is far more serious, I have to dig into the code now. Or perhaps try GNV?

(Who called this "Open" ? Those Unix guys, propably. If it were really open, I wouldn't have this trouble building it)