Tim Janik wrote:
i think the direct comparison in question was written for efficiency reasons. a quick test shows, that, depending on the value of v_identifier, introducing strcmp() makes the current code from 1.5 to 2.5 times slower, introducing g_strcasecmp() makes the code around 7 times slower. that's with glibc and -O2, note that glibc can already unfold certain strcmp() calls, while the comparatively bad performance of g_strcasecmp() is due to GLib rolling it's own loop.OK to commit?nope ;)
I think that the explanation above belongs into the code as a comment. :) Miro