If we coerce from, say, R = QQ['x', 'y'] to S = QQ['y', 'x'] than R(f) (a,b) = S(f)(b,a) which I think could be confusing, so it might take some convincing to make me think that coercing from R to S is a good idea. For the strict subset case, the calling conventions will have a different number of arguments, so it would be less "surprising" (worst case an error rather than a bad result).
- Robert On Mar 29, 2007, at 8:56 PM, William Stein wrote: > On 3/29/07, David Roe <[EMAIL PROTECTED]> wrote: >> I guess I'm not totally sure what you meant by (3). Do you mean >> (3) If (2) is satisfied, then for all v variables of R and w >> variables of S that are not in R, v < w? > > Case (3) is *only* invoked in case the set of variables are equal > as sets (but the order could be very different). In that case, > it's the tie breaker to determine in which direction to coerce. > >>> Actually, whatever we do, we should make sure and require that >>> the term orders on the two rings are compatible. >> What is the condition for compatibility? > > I don't know, but we should probably think about it. All the > multivariate > polynomial rings in SAGE are equipped will a term order. If we > have two copies of QQ[x,y], but with different term orders, in > which way > does the coercion map go? My point is that there is more to > coercion > than just the variables -- the term order is also perhaps relevant. > My inclination is simply to never do automatic coercion if the term > orders aren't "equal" in some sense (where equality could be > determined > by the TermOrder class). > > William > > --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~----------~----~----~----~------~----~------~--~---
