Front page | perl.perl5.porters |
Postings from January 2012
Hi all, The changes outlined below are now in blead. You know where to find me for sending bombs. On 01/30/2012 08:32 AM, Steffen Mueller wrote: > Before the big feature freeze, I plan to merge (well, > rebase-rebase-push) a moderately large set of changes to the XS and > typemap documentation. I set out to document all undocumented and test > all untested core typemaps a while ago and the aforementioned changeset > is the result of that effort. Here's what I got: > > - Documented and tested several more typemaps. > > - Found two typemaps that apparently aren't used anywhere in core or on > CPAN. Their purpose is undecipherable. I want to remove them from the > core typemap. More on that below. This not done yet, but it's a trivial change. I've laid the groundwork in that the newest ExtUtils::ParseXS release on CPAN (dev release so far!) comes with all the necessary tools. The ExtUtils::Typemaps::Excommunicated module for these goners is ready for upload to CPAN, too. > - Moved typemap related documentation from perlxs.pod to a new document, > perlxstypemap.pod, which goes into much more detail and contains a list > of all core typemaps with their purpose (except for those that I haven't > managed to cover yet). Very importantly, the documentation contains a > list of Perl variables that are useful for interpolation into typemaps, > courtesy of Tony Cook. Included. > At the same time, I continuously applied changes to > dist/ExtUtils-ParseXS in blead to add better facilities for typemap > reuse. This means that it's now rather easy to share typemaps between > CPAN modules. I've wanted to do this for some time now, but my plan to > rid us of some of the WTF typemaps above has made that a necessity. > Here's the plan for getting rid of them: > > - I release ExtUtils::Typemaps::Excommunicated to the CPAN that contains > T_DATAUNIT and T_CALLBACK (the latter does not do what you think it does). Will do after the next stable ExtUtils::ParseXS release (which happens whenever I get to it). [...] > I see no good way to have a proper deprecation cycle for these typemaps > that isn't at least as invasive (adding a mandatory warning to the > generated code would be as obnoxious, IMO). Let me say again: Not found > on CPAN. Bizarre code that seems to have served a very specific purpose > lost in the midst of time. > > This changeset is now a smoke-me branch as > smoke-me/smueller-typemapdocs7. Please speak up if you oppose to any of > the above. Too late to prevent me pushing this. --SteffenThread Previous | Thread Next