On Tue, 2010-01-12 at 03:52 -0500, Behdad Esfahbod wrote: > On 01/12/2010 03:44 AM, Philip Withnall wrote: > > On Tue, 2010-01-12 at 03:38 -0500, Behdad Esfahbod wrote: > >> On 01/04/2010 12:53 PM, Javier Jardón wrote: > >>> Hello, > >>> > >>> The objective of the GnomeGoal is to add code coverage of your code > >>> with GCOV [1] > >>> You can take a look to the GnomeGoal page here: [2] (There are also > >>> examples to add GCOV support to json-glib and libgdata) > >> > >> I was a huge fan of adding GCOV support to all modules, and actually have > >> patches for glib/pango/gtk+ in bugzilla. But after finding out that coverage > >> can be extracted without modifying the modules (like build.gnome.org is > >> already doing), I'm against cluttering the modules with GCOV boilerplate. > > > > Having the gcov stuff in-tree means that you can test the coverage of > > new code as you write it, so you could theoretically always have 100% > > coverage of the code which is committed. > > We've had that code in cairo for years. But getting coverage is so slow that > I doubt people use it regularly. Surely it's only as slow as running the test suite? > > Similarly, build.gnome.org only tests the versions of modules in the > > current jhbuild moduleset, not the latest git versions. > > > > It's not much boilerplate code anyway, so I thought it was useful for > > libgdata. To each their own. :-) > > Certainly disagree with the "not much boilerplate code anyway". Looks like a > 100 lines of copy/pasted code to me. That's where all the cruft comes from... 50 lines, but it could easily be moved to gnome-common. bgo#606720 filed. Philip > behdad > > > Philip > > > >> behdad
Attachment:
signature.asc
Description: This is a digitally signed message part