axiom-developer
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Axiom-developer] Re: Which source distributions should we list?


From: daly
Subject: [Axiom-developer] Re: Which source distributions should we list?
Date: Sun, 27 May 2007 23:41:16 -0500

Ralf,

> Which source distributions should we list?

I believe we should list them all. Each developer has a choice of the
source code control mechanism that fits his needs and interests.  Gaby
and Waldek seem to like SVN. Bill likes Darcs. The silver version uses
SVN/git. The main gold version will continue using the ARCH/CVS tools.

The only "end-user visible" systems should be the ARCH/CVS versions.



> I still don't see a good structure. 

A good structure is related to your style of code development.  Which
system do you use to maintain ALLPROSE? I haven't seen a branch off
any of the existing systems but I'm assuming you chose one.

Having used half a dozen systems (SCCS, SourceSafe, RCS, Arch, CVS,
SVN) in the past I can safely say that git is my choice. 

I won't even bother to list the many reasons except to say that it
has completely changed the way I work and the way I think about how
to develop. Only Lisp and Emacs have had a similar profound impact.

I know these things border on religion and we really don't need
any more debate about it. Choose your system. Use it. As long as
the changes get diffed against Gold it doesn't really matter.



re: PATCHES

> I somehow don't like to create and sent patches.

The stated practice of the project at the moment is to post
diff -Naur patch files which are accepted by all the systems
and easily manipulated. It's not hard; given a change type:

diff -Naur originalFile changedFile >originalFile.patch

and post the resulting originalFile.patch.

Just be sure that the originalFile is the one from Gold or
silver so that everyone has a baseline from which to start.
Otherwise you run the risk of changes that not everyone can
apply to their system. This is precisely the problem that 
touched off the Makefile newline debate.

Linux lived and grew using this method despite having hundreds
of developers using many different systems. It is suboptimal
but it works everywhere.

t











reply via email to

[Prev in Thread] Current Thread [Next in Thread]