RE: Hmmm.... future of cons?

[Steven Allen]

Just my 2 cents..

Oddly enough, there are only two reasons that the big company I am
contracting at now allows the use of Cons in a small part of our project.

1.  It's perl - people here know perl, and use it to do things that were
previously very painful to do in Make.  People here write lots of perl
scripts.  *No one* except for me knows python.

2.  Signature checking - this gets around the time stamping issue with many
machines, etc...  (We've debating on editing gnumake to add signature
checking).

Despite everything that Cons does, there are the *only* two things that
management here cares about.  So if Cons becomes stable, by release 2.3.1
and possible 2.4 in the future, people here may slowly convert our make
system to cons (though with 20841 files at last check - that task alone
would take months and months).  But doubt it.  Cons is still a fringe tool
in many people's eyes, and without a large community using it and fixing it,
they probably won't use it.  Yes, it's only a perl script, but it ain't
small and while it's probably a fine design, I can't figure most of it out,
so I'm not gonna ask people here to do that.

Moving to SCons is *not* going to happen here, period.  Personally, I know
perl and python, and like them both for assorted reasons.  But when I leave
at the end of my contract,  Cons' future here is dubious.  SCons has no
future here purely because it uses python.  People know make, it sucks, but
they know how it sucks.  Local/in house expertise on a bad tool (make) will
win out over a good tools (Cons) with questionable support from a
disintegrated internet group.

Personally, I will switch to Python/SCons because I like it's goals, and
extending SCons is important to me.  I work on projects that call dozens of
specific tools, so ease of extending to useful.  But until SCons is stable,
I'll have a hard time selling it.

Hmmm, I hope that wasn't too rambly...

Steve

-----Original Message-----
From: H. S. Teoh [mailto:address@hidden 
Sent: Wednesday, May 22, 2002 10:06 AM
To: address@hidden
Subject: Re: Hmmm.... future of cons?

On Wed, May 22, 2002 at 09:32:27AM -0400, address@hidden wrote:
[snip]
> This brings up the question of what "moving on" means for those currently
> using Cons.  Is/will there some way to move from Cons to SCons/Python?
> Other than rewriting all Construct and Conscript files by hand I mean.
[snip]

Exactly. I have moved several of my personal projects to Cons. Why do I
have to revisit all of them again just to re-write the conscripts for
SCons? Plus, I have started using Perl scripts for various aspects of the
build (although not part of the Conscripts themselves) -- moving to SCons
means my projects now have yet another build requirement (Python). I
wouldn't want to rewrite all those Perl scripts in Python just because I
want to use SCons.

Maybe, as somebody pointed out, we should have a core Cons/SCons package,
written in whatever language of choice the developers happen to like, and
then have front-ends that can work with either Perl or Python (or whatever
else people might want, like Ruby). If designed properly, the front-ends
should be relatively simple to implement for various different languages,
and we don't have to duplicate effort implementing the core. If SCons
could go this way, then I'll happily support it and keep my existing
Conscripts in Perl.


T

-- 
WINDOWS = Will Install Needless Data On Whole System -- CompuMan

_______________________________________________
address@hidden
http://mail.gnu.org/mailman/listinfo/cons-discuss
Cons URL: http://www.dsmit.com/cons/