On Fri, Dec 10, 2010 at 1:18 PM, Simon King <simon.k...@uni-jena.de> wrote: > Hi Robert, > > On 10 Dez., 19:51, Robert Bradshaw <rober...@math.washington.edu> > wrote: >> As for the more general question, the current state of comparison of >> elements in Sage is a huge mess, and not much should be read into the >> current structure. It was written before cpdef existed and before the >> interaction between __cmp__, __richcmp__ and __hash__ was fully >> understood, and then more hacked on rather than fixed with every >> iteration of the coercion model. It really just needs to be fixed for >> once and for all (e.g. make a system where elements define a cpdef >> _cmp_ and _hash_, > > ... and _richcmp_, if one wants cmp(.,.) to behave different > from .==., right?
Yes, and our _cmp_ and _richcmp_ would coerce for you, like all the other binary operations. Wow, this is motivating me to finally do something about it... > In particular, it would be nice if the methods to be implemented were > the same for Python and for Cython classes. Yes, for sure, hence cpdef. >> with a default __cmp__, __righcmp__, and __hash__ >> provided, and anything needing to do something more fancy (e.g. >> symbolic elements) define their own __richcmp__ etc. doing coercion >> themselves (the binop function is simple enough to use here). There >> was an attempt made as part of the penultimate coercion re-write, but >> it was too invasive on top of everything else that was changing at the >> time. > > Yeah, I can imagine that it is hard to do such a big change. > > Thanks to you and Jason! Thanks to you for helping push this through to completion! I can't wait until we can finally delete parent_old and all the associated ugliness. - Robert -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
