On May 14, 10:50 am, Tim Abbott <tabb...@mit.edu> wrote:
Hi Tim,
> Sorry for the slow reply. sage-devel's Reply-To munging drops
> everyone from the Cc: on replies and I'm normally not directly
> subscribed to the list.
I tend to use Google groups directly and not bother with the email
interface.
> (because of this kind of problem, I do think that everyone should run
> their high-traffic mailing lists without any sort of reply-to munging,
> like the Linux folks do)
>
> On May 6, 9:46 pm, mabshoff <mabsh...@googlemail.com> wrote:
>
> > > Since they haven't released in several years, this might be a good time to
> > > send them any patches Sage has against ghmm that haven't been applied
> > > there already.
>
> > Well, William did some pretty deep changes upstream does not want,
> > i.e. Sage does not use the XML interface that they want to keep non-
> > optional, so I am not sure how much of the code should actually go
> > upstream.
>
> Yeah, that does sound potentially problematic. What is the motivation
> is for avoiding the ghmm XML interface?
To minimize the dependencies. I am not too familiar with the details,
but William considered the dependency on some XML library too heavy
for Sage.
> Is the plan for Sage to maintain the changes to evade the XML
> interface indefinitely?
Well, I believe so. I do think that in Debian linking against a clean
upstream ghmm would "just" require to link some extra libs into the
extension, so that sounds doable.
> Does the Sage patch for this at least not break any functionality in
> upstream ghmm? If so, I could potentially include it in the Debian
> ghmm package.
I would consider the thing that Sage does breakage of upstream, so the
above option might seem better.
> > > I can't figure out what those are, however, because the .spkg in Sage
> > > 3.4.1 seems to have a bunch of patches that don't do anything (i.e. the
> > > file being copied in is replacing an identical file). Did someone
> > > accidentally overwrite the copies in the upstream directory, or did the
> > > patches all get merged already?
>
> > Sounds like an accident.
>
> Yeah, that's what I figured. It'd be helpful if you could shoot me an
> email when that gets cleared up so I can take a look at the Sage
> diffs.
Sure. William will need to sort that out :)
> > As is the code in Sage is actually partially
> > broken on Itanium/Linux (some allocator just segfaults), so this ought
> > to be sorted out in Sage 4.0.x. Right now the plan is to do 4.0 before
> > SD 15, 4.0.1 after SD 15 with work from SD 15 and patches that didn't
> > make it into 4.0. and then hopefully sort out everything for Debian
> > for 4.0.2, i.e. about a month from now. Does that fit with your time
> > table?
>
> > After that we ought to have 4.1 which should deal with the category
> > patches, etc.
>
> That schedule sounds reasonable to me. To be honest, I'm incredibly
> busy right now with my startup (Ksplice) and so I can't make any
> commitments regarding when I'll find time for Sage work.
Sure. As mentioned on my end you have the commitment that Sage 4.0.2
will unless something really bad happens ship upstream releases all
the way and be current with whatever upstream has released. After that
the pari-svn transition might happen for example, but it is unclear,
so either way Sage 4.0.2 might be a good candidate for packaging in
Debian even if 4.1 is out. The 4.0.2 planned release date is now June
13th, 2009, i.e. shortly before MEGA 2009 in Spain which willbe
followed by SD16.
> -Tim Abbott
Cheers,
Michael
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to
sage-devel-unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://www.sagemath.org
-~----------~----~----~----~------~----~------~--~---
