Here is the root problem, which I think illustrates some unintended consequences.
Changes at #10627 make possible things like printing a matrix with -, 0, + if the entries are negative, zero or positive. Which is a very nice thing to have. You can also specify a translation dictionary between values in a matrix and any alternative you would like to print. Making this work requires a look-up in the translation dictionary, which will form a hash for each matrix entry. Now, for entries from QQbar, these hashes can be extremely time- consuming - like a full minute. (This is #11543.) So this feature really has a harmful effect on the performance of printing matrices. Furthermore, for QQbar, it appears the hash adjusts the precision of the QQbar entries, so the act of printing makes a small change in the entries, which is very evident for entries close to zero. Irony is, I was the reviewer on #10627. Patch now ready to go at #11544 simply by-passes the look-up in the dictionary when the dictionary is empty. (Right now there are lookups in an empty dictionary.) So it avoids many problems for "routine" matrix printing, but is not a full solution. Still it would be a big improvement for the usefulness of matrices over QQbar. It is literally a change to one line of code, and then a bunch of very minor doctest adjustments. Hint, hint. http://trac.sagemath.org/sage_trac/ticket/11544 -- 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
