Re: [Axiom-developer] Pamphlet format discussion

[Bill Page]

On 21 Jul 2007 03:29:01 -0500, Gabriel Dos Reis wrote:
> "Bill Page" writes:
>
> [...]
>
> | Integrating noweb into Axiom to support the current pamphlet file
> | format seems entirely reasonable to me. I think calling noweb as a
> | system supplied utility is the best approach.
>
> Please don't go out of your way to "copy" noweb into Axiom.  Make it
> a dependency.

Yes I agree. noweb should *only* be a dependency *not* copied into the
Axiom source. It is a tool to be used just like gcl uses gcc - not
more.

> At IBM Toronto, where we just had an ISO C++ meeting, I spoke again
> with Benjamin Kosnik (from Red Hat, in CC:) about what would remain to be
> done to get Axiom packaged and distributed in Fedora Core (for
> example).  I explained what has been done in terms moving to standard
> build system.  I pointed out that at this moment we still critically
> depend on GCL and noweb.  He quickly made a search and informed me
> that FC distributes GCL-2.6.7-something (but I guess the next FC may
> distribute GCL-2.6.8 if it is finalized by then), but they do not
> have noweb at the moment.

That is excellent. Thank you for bringing this up with him.

Perhaps you will recall that Camm Maguire actually distributes the
current gcl-2.6.8pre as a "fix" release to gcl-2.6.7 on Debian.
Perhaps a similar approach should be taken with Red Hat?

> He looked at the noweb page and suggested
> that a first step would be to list the noweb RPM as a dependency in yum
> package and see how it works.

Could you explain what you mean by this? What yum package? Do you mean
that we shoulld encourage the noweb developer (Norman Ramsey) to
submit such a package?

>  Benjamin is very willing to provide
> advice on how to make forward progress with these issues.
> I have not spoken with the SuSE people recently about the progress we
> have made with Axiom.build-improvements, but I would be surprised if
> they didn't say roughly the same thing.
>
> We should try hard to avoid complicating system packager's life.

Indeed! We want to make it as easy as possible to re-distribute Axiom.

Regards,
Bill Page.