[10:05] morning [10:15] uploaded -intel 2.20.19 with sna [10:17] weird, so libradeonsi.a is suddenly 27mb, but each individual object summed together is much smaller [10:19] oh, links to libradeon [10:24] we're going to drop it from raring anyway, would it be uncool to drop from quantal too? [10:43] what do you mean? [10:43] radeonsi is disabled in my mesa snapshot [10:44] because the llvm backend was moved to llvm, and will be released with 3.3 [10:44] oh that [10:46] I'd rather keep the quantal lts stack's support identical to quantal. I was thinking of trying to move everything in drivers/* to libgallium, to see if it reduces the object size some. I wonder if llvm's static libs are duplicated multiple times.. [10:47] sure [11:08] hm doesn't look easy, I'll try an evil hack [11:08] just building llvm shared from the static files it has [11:53] didn't seem to affect size much, though :( [11:53] hrm [11:58] I guess everything would already be pulled in from -lLLVM-3.1 [12:01] also reduces the possibility of making it even smaller :( [13:27] so, now I actually uploaded -intel.. [15:07] phew, didn't do a reboot in my chroot this time, almost did though :P