Re: extra release step

() Raimund 'Raimi' Jacob <[email protected]>
() Sat, 28 May 2011 12:44:52 +0200

   While I would pretty much like serveez to be run under windoze - this seems to
   be the point where we get rid if that port. For the following reasons:

   - win32 is actually 2-3 platforms: cygwin, mingw, native (Visual Studio)

Thanks for explaining this; i had an improper mental model of woe32 previously.

   a proper port would be mingw or native, while cygwin would probably be an
   easier port. but cygwin is a PITA performance-wise and only a big work-around
   in many aspects. A serveez port would be of limited use.

   - there is currently no one with a build system (for any windoze
   port). neither Stefan nor me (nor probably Martin) even have a windoze system.

   - there is no one asking for the windoze port.

   - guile is not officially supported, as Mike pointed out.

   - the entire source becomes easier to read/maintain without the cruft.

   There is only one argument for keeping the windoze port: The educational
   aspects of having portable network programming. hm.

Upon reflection, this last argument weighs (perhaps disproportionately)
heavily on me.  I cannot forget that if it had not been for my introduction to
GNU Emacs on a proprietary platform (VAX/VMS ~1990), i would not be hacking
Serveez now ...

   I say: Drop it and let the unicorns dance with the singing birds.

... so, i take back my previously expressed culling glee and propose an
intermediate course: drop "native woe32" and continue "support" for cygwin
and mingw.  Removing build-aux/vcgen.bat weirdness is enough for me.  With a
few months of experience with the other woe32 bits, i can say it's not such a
big deal to maintain.  What is a good platform but a bad platform's teacher?
What is a bad platform but a good platform's job?

Is this course ok w/ everyone?