The new tuxracer ebuild won't merge for me. I think this is because of some sed magic in the ebuild. It replaces -malign-jumps, etc. with -falign-jumps, etc. in the Makefile. This is fine for gcc 3.x, but not for gcc 2.95.3. (At least, this is what the gnu docs seem to suggest.) So, I hacked the ebuild so that it would only do the sed trickery when the user is running a gcc 3.x system. I also took out the two extra "cp Makefile Makefile.orig" statements, since that would overwrite the Makefile.orig w/the sed-ified version. Correct me if I'm wrong. Let me know if I'm completely off base here...the fix seemed to work for me though.
Created attachment 4419 [details, diff] the aforementioned patch
I can second this. Tuxracer only builds here on my gentoo 1.2 with gcc 2.95.3 when i comment out the handling of these new compiler options for gcc 3.x Seems to be a sad tendency for gcc 2.95.3 users that new ebuilds and ebuild updates are only getting gcc 3.x tested :(
yeah these updates were kinda sucky huh ... ill fix them when i get back to AXP and when the network stops sucking ;)
the multiple sed cmds are correct, they work the way they're supposed to its just kind of prettier to use `sed -e 's:::' -e 's:::' -e 's:::'` ...
just committed the fix to cvs ... re-open the bug if it doesnt work for you
Yea, I was kind of unclear. I agree that the multiple sed statements work. I don't think that you want the two extra "cp Makefile Makefile.orig" statements though. (The ones with the arrows pointing to them.) cp Makefile Makefile.orig sed 's:-malign-loops=2:-falign-loops=2:' \ < Makefile.orig > Makefile ---> cp Makefile Makefile.orig sed 's:-malign-jumps=2:-falign-jumps=2:' \ < Makefile.orig > Makefile ---> cp Makefile Makefile.orig sed 's:-malign-functions=2:-falign-functions=2:' \ < Makefile.orig > Makefile The Makefile.orig will no longer contain the original Makefile. It will contain the output after the first two sed expressions are run on the Makefile. This is is not a big deal. It seems that you intend to back up the original Makefile though, so these two extra cp statements would screw that up. Correct me if I'm wrong..
The new ebuild works for me. Disregard that stuff about copying the Makefiles. The new ebuild doesn't have that problem. Thanks a lot.
the cp statements are needed ... (legend) -al ==> -malign-loops=2 -aj ==> -malign-jumps=2 -af ==> -malign-functions=2 cp: Makefile (-al -aj -af) Makefile.orig (-al -aj -af) sed: Makefile (-aj -af) Makefile.orig (-al -aj -af) cp: Makefile (-aj -af) Makefile.orig (-aj -af) sed: Makefile (-af) Makefile.orig (-aj -af) cp: Makefile (-af) Makefile.orig (-af) sed: Makefile (all replaced) Makefile.orig (-af) final Makefile has all 3 flags replaced ... if you removed all but the 1st cp, you'd get this: cp: Makefile (-al -aj -af) Makefile.orig (-al -aj -af) sed: Makefile ( -aj -af) Makefile.orig (-al -aj -af) sed: Makefile (-al -af) Makefile.orig (-al -aj -af) sed: Makefile (-al -aj ) Makefile.orig (-al -aj -af) final Makefile has only -af replace ...
My mistake. You're completely right. It won't work without the two cp statements. Sorry about that. I probably shouldn't be writing bug reports when sleep deprived...
no reason to apologize all the time you dont learn anything without asking questions ... i would rather have people ask more questions than never ask any for fear of sounding stupid
