New maintainer for EMX?

Ilya Zakharevich <nospam-abuse-qlYHBhoNh+/[email protected]> Wed, 21 May 2003 15:01:44 -0700 (PDT)
Newsgroups gmane.comp.ide.emx.devel
Organization U.C. Berkeley Math. Department.
Message-ID <[email protected]>
[Posted ot comp.os.os2.programmer.misc and Cc'ed to EMX mailing list.]

Can't we find a new volunteer for maintainance of EMX.DLL and
EMXCRTM.DLL?  The job is to *collect* contributed patches, weed out
the weeds, and decide when we have a snapshot good enough to be named
golded.

I do not think the person should be very knowledgable about OS/2
internals; I would think a very good integrity *is* required, as well
as good skills in patching, changelogging, documenting - or at least
an ability to force the patchers to do all the changelogging and
documenting.  ;-)

Myself, I can't be a maintainer.  However, I promise to contribute the
following:

  a) a fix for a hang when starting asyncroneous processes after a PID
     wrap (ready);

  b) a fix for OMF executables not being able to fork() (prototype ready);

  c) a fix for needing all these -Zso etc switches for DLLs (prototype ready);

  d) a fix for not being able to handle serial port speeds except the
     few values such as 1200,...,38600 baud;

  e) a fix for not being able to PASTE more than circa 1000 bytes to
     TERMIO(S) applications;

  f) a fix for socket handles not being inherited by childs;

  g) 4 or 5 other minor fixes may be quickly migrated from Perl's os2.c.

I also may be able to fix a couple of bugs in sprintf(%g) I know.  I
can't promise to do a fix for being confined to "low memory" only -
simply because I do not have a post-Warp3 system.

So: anybody volunteering?  EM says that only circa 500 bugs are known.
Assuming that 10% appear in 90% of cases, I think that taken together,
we can fix most of the problems during the summer.  Having a good
quality of fixes *is* a problem - EM's code looks very clean, and we
must try to do not much worse.  Running all the patches through the
mailing list and many people actually *reading* them may help.

The actual problem is that EMX comes without a viable test suite;
however, for what I know Perl test suite + pdksh test suite should
cover about 50% of that; maybe Python can add something too?

Judging by the license, we can do not *need* EM's consent - although
his help and consent would be crucial...

Ilya