On the plus side for OpenVMS folk, VSI seems at least to have rescued
it from total extinction.  On the minus side...   well, it's still its
own beast of a niche system.  And while it's not yet dead, and
probably won't be for quite some time, I don't think it's existence in
a X11 workstation role is a large percentage of its market presence
(or that that's likely to change in an upward direction).  So, I
propose to retire our pretensions of support for it.


ctwm has historically had some level of VMS support, often maintained
by sheer stubbornness on Richard's part ;).  However, it's different
enough from the rest of the world that we'd pretty much need somebody
regularly running it to be sure the code kept working.

And it has also used a completely separate build infrastructure (and
separate pre-generated files for lex/yacc output, etc), which means it
couldn't even build to check the code unless that was kept up in
parallel too.  Which it hasn't been; I'd lay good odds 3.8.2 couldn't
even be built without extensive reworking, and I suspect nobody's even
tried since maybe 3.8 at the latest.  And current head is leaps beyond
that.


Why bother removing it, as opposed to other ancient platforms we're
just quietly allowing to rot?  Because it's enormously invasive
compared to them; huge blocks of #ifdef make the code harder to read
and search through, and there are a lot of other files hanging around
purely devoted to it.  In preping the retirement, I deleted 19 files
completely, that never had anything to do with anything but VMS, in
addition to blocks of code removed from the other files.  A little
wc(1)-age shows a bit over 250k removed; that's more than a tenth of
the full dist (including a meg of XPM; count that out, it's closer to
a fifth).

What's the downside?  Well, if somebody ever does want to revive it, a
lot of the #ifdef'd code would presumably be needed.  I suspect at
least some of it and the extra files would be historical cruft even
then, but hard to know from where I sit.  Having it removed from the
current code means anybody wanting to bring it back to life would have
to trawl back into the VCS history to recover them, or manually redo
it all.  Not ideal.  OTOH, it being apparently there but thoroughly
rotten may not really save them all that much effort, while adding to
the load of the rest of us until then.  And...  well, we HAVE VCS
history for those sorts of things.


So, between the cost of carrying it, the apparent recent and current
non-use, and the lack of expectation of that changing, I think the
balance is pretty heavily in favor of retiring it.  I'm willing to be
convinced otherwise; certainly it's a nice selling point we have
against...  like, every other WM on the planet (though maybe that says
something about its utility).  But in the absence of somebody actively
using and maintaining the platform, I think it'd take some pretty hard
convincing.

What say we?


-- 
Matthew Fuller     (MF4839)   |  [email protected]
Systems/Network Administrator |  http://www.over-yonder.net/~fullermd/
           On the Internet, nobody can hear you scream.

Reply via email to