Am Dienstag, 19. September 2006 22:05 schrieb Christian Ohm: > I tested waf a bit and it seems to work well. I had some of those "can't > link lots of {[]w9r" at the beginning, but that didn't occur later (at > the latest with your second version). When you run waf with --nocache it doesn't seem to produce those crappy LIB_XXX configs. Seems like the caching engine sometimes (randomly?) confuses the LIB_...s or LIBPATH_... and stores a whole structure in them. I bet when I report that now it will be fixed next week or even tomorrow. ;)
That's what I love about waf and Thomas. I told on the mailinglist that the current way of configuration confuses me and a week later he had reimplemented the whole config (with help and tips from others of course). :) > Speedwise, my first test was blindingly fast compared to the make build, > but that was because '-O2 -g' was missing from the cflags. With those > options, both are almost the same speed. The build is probably limited by the speed of the compiler. But the configuration is blindingly fast. > I noticed two things: > > 1. On the first run, waf scans all source files (for dependencies), that > takes almost 20 seconds here with no output or indication on what's > happening. Forgot to tell you about it. Sorry. (And I will tell Thomas that he should give a progress info.) > 2. The waf script contains the whole wafadmin directory as a base64 > encoded tar.bz2 archive. That means updates to waf will exchange the > whole base64 part, as a text diff, and not only the parts that actually > changed. We might want to have waf-light and the wafadmin directory in > svn and just make the waf script for releases, that'd solve that > problem. Yes, but as I found it easier to send the complete waf script to this mailinglist, I just did it this way. > No showstoppers in my eyes, merely small annoyances. > > Oh, and the -p option gives an incentive to eliminate the warnings :) When you have a look at the first waf script I sent, then you'll see that the warnings don't appear there... He changed it from capturing the stdout/stderr to not do it because of bugs in Python when there is lots of output (in this case when there are lots of warnings). --Dennis
pgpiQmbFy0zHZ.pgp
Description: PGP signature
_______________________________________________ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev