axiom-mail
[Top][All Lists]
Advanced

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

Re: [Axiom-mail] Compiling Axiom on FreeBSD


From: Mark Murray
Subject: Re: [Axiom-mail] Compiling Axiom on FreeBSD
Date: Sun, 09 Nov 2003 20:23:11 +0000

root writes:
> Mark,
> 
> The diffs would be best, actually. Can you diff them using:
>   diff -Naur oldfile newfile >newfile.patch

Er, can I do a "cvs -q diff -udN > file.patch" instead? :-)

> In general, I don't fork other packages. The scheme is to pick a
> particular version (normally the latest if it works) and put a
> tar-gzipped copy of the sources in the zips directory. Make untars,
> applies Axiom patches, configures, and makes the other packages.

(I thought we had this discussion before?) I seem to recall that
you were planning to extend noweb?

Having the package as a non-severable bolt-on-the-side makes the
porting process much harder. FreeBSD already has a functional
noweb port, and the noweb that comes with Axiom is broken, so we
either need to patch noweb (done at the moment on the belief that
it was going to be forked), or just cut off, which is much preferable.
Would it be possube to make the cutting off option easier?

Another example, the GCL that you have is completely non-functional
for FreeBSD, without a rather large patch set. OTOH, FreeBSD has
a perfectly functional port of GCL-2.5.3, and 2.6.1 and 2.7.* are
on the way.

In order to build for FreeBSD, I had to work quite hard to make the
build happen, and I still haven't gotten it clean.

> This gets around a lot of issues as it makes the original sources
> available, makes clear what Axiom needs to patch, and makes sure
> that the packages run on the current install since they are built
> from sources. 

I hear waht you are saying, in that it does make certain things easier
to track, but it makes porting issues much more complicated.

> As I find things to patch I'll generally contact the maintainer of the
> packages and ask them to accept them. Some patches are so Axiom
> specific that it makes no sense and some are just not something the
> maintainers would ever want. So the Axiom versions are not forks in
> any traditional sense as we are trying hard to track the standard
> packaage builds and keep up to date. For example we just moved from
> GCL-2.5.2 to GCL-2.6.1 by replacing the tgz, removing accepted patches
> and porting Axiom specific patches.  It all went rather smoothly.

The GCL issue broke me :-)

> For noweb I do the same thing. I'll apply your patches to noweb's
> makefiles before I execute them. You'll see the patches show up
> in the zips directory with names ending in .patch. These will be
> applied (and explained) in the top-level Makefile.

If there is no plan to fork noweb, then it makes much more sense for
me to force use of a local port (already live and well).

M
--
Mark Murray
iumop ap!sdn w,I idlaH




reply via email to

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