I just talked to Jürg in IRC, and he prefers the old version too, so it's going to get reverted. Jiří, do you think the attached patch is sufficient for documentation? It just adds a link to l.g.o/Vala/MarkupSample, which we can change at will. I don't think we should put too much documentation in the vapi, but I like the idea of at least linking to a sample... -E On Tue, 2010-01-26 at 23:28 +0100, Jiří Zárevúcky wrote:
Michael 'Mickey' Lauer píše v Út 26. 01. 2010 v 23:24 +0100:Am Dienstag, den 26.01.2010, 15:48 -0600 schrieb Sandino Flores Moreno:The next URL has the testcase: https://bugzilla.gnome.org/attachment.cgi?id=152347 It demonstrates that using methods as callbacks with MarkupParser is possible without introducing strange parameters neither restricting the methods to be static. It's the same parser I had written in Genie, just simplified and translated to vala.Thanks for this example. This indeed convinces me to toss a vote for the revert -- sorry Jiří. In this case I prefer usability over theoretical correctness.Theoretical correctness aside, any other use case leads to broken C code, which is far from my definition of usability. I have no problem with the old way, *in case* it gets properly addressed in the documentation *and* as a comment in the VAPI. There is nothing more annoying that bindings that require undocumented magic to work.
Attachment:
0001-glib-2.0-revert-previous-patch-against-MarkupParser-.patch
Description: Text Data