Hi John, On 2015-09-12, John H Palmieri <jhpalmier...@gmail.com> wrote: > By the way, I think the old database is broken: I think it tries to install > into some directory which now no longer exists. So I think the one in the > newest beta is the only option.
In the same way (changed directories, changed import locations) version 2.1.3 and 2.1.4 of my group cohomology spkg got broken. I was told that changing these directories was an *internal* change and thus did not require a deprecation period. By other internal changes (Cython upgrade, Singular upgrade), versions 2.1.1 and 2.1.2 broke as well. In other words: All versions from 1.0 till 2.1.1 were motivated by improvements. All versions from 2.1.2 till now were at least partially motivated by coping with backward incompatible changes. For the record: I think ANY backward incompatible internal change that *can* be covered by a deprecation warning (thus, changed import locations, changed function names or argspecs, etc) *has* to be covered by a deprecation period, in particular when it breaks some optional package. Admittedly, internal changes in third party packages (cython, singular) do not fall into that category. I thought some patchbots did test optional packages on a more or less regular basis? Anyway. Isn't it very convenient that one can say: "The old-style optional packages are now all broken (by internal changes that did not require deprecation), thus, let's expunge them." Cheers, Simon -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
