On Wed, Jun 02, 2010 at 02:55:39PM -0700, William Stein wrote:
> On Wed, Jun 2, 2010 at 1:18 PM, Nicolas M. Thiery
> <[email protected]> wrote:
> > There is consensus that no-one cares about oldish crystal pickles,
> > and it would make the inclusion of #8911 technically easier if we
> > could drop them.
> >
> > Hence my question: what's the official procedure for updating the
> > pickle jar?
>
> I propose the following official position:
>
> "Don't do it, unless there is a consensus that doing so won't
> cause serious concern. Moreover, any change in pickling that replaces
> a pickling system by a new one, should be *designed* to be backwards
> compatible, in case future changes are proposed. See, e.g., the
> matrix and vector code, where the pickling is all designed from the
> start to allow for new pickling formats without breaking old ones."
Sounds good to me.
> Regarding your changes, have you re-architected the code so your new
> pickling model is backwards compatible?
Yup, that's one of the benefit we got from switching to
UniqueRepresentation; and that's precisely this switch that broke
backward compatibility (not that it was infeasible, but it would have
taken some work). All of this essentially boils down to the fact that
the crystal code was the first one we wrote in Sage, and we did not
have the proper infrastructure at that time.
So now, from a technical point of view, what's the procedure for
updating the pickle jar? Should I use the SAGE_PICKLE_JAR feature
advertised in sage_object, run the tests on sage/combinat/crystals,
update /extcode/pickle_jar/pickle_jar.tar.bz2 with the new pickles,
and post this on the #8911 ticket?
Speaking of the pickle jar. I did not get comments on my e-mail about
#9075. It suggests that quite a few objects in the pickle jar might
actually be broken, even if they don't raise an error upon unpickling
(that was the case for our crystals).
Best,
Nicolas
--
Nicolas M. Thiéry "Isil" <[email protected]>
http://Nicolas.Thiery.name/
--
To post to this group, send an email to [email protected]
To unsubscribe from this group, send an email to
[email protected]
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
