brooks@maddog.llnl.gov (Eugene Brooks) (12/15/88)
Wanting to get some serious response speed out of X11R3, I fetched the PURDUE speedups and the PURDUE+ speedups from expo.lcs.mit.edu. These patch files are causing quite a few .rej files with patch. The PURDUE speedups caused one .rej file which I managed to get by on with some brace format changes to the pattern it was looking for, the PURDUE+ speedups are so full of rejected patches that I don't dare fix them all by hand. What happened here? Is anyone else getting nailed heavily by the PURDUE+ speedup patch kit? If the patch files are bad, could some kind soul place useful patch files for the PURDUE+ speedups on expo?
rusty@garnet.berkeley.edu (12/15/88)
In article <14671@lll-winken.llnl.gov> brooks@maddog.llnl.gov (Eugene Brooks) writes: >Wanting to get some serious response speed out of X11R3, I fetched the >PURDUE speedups and the PURDUE+ speedups from expo.lcs.mit.edu. These >patch files are causing quite a few .rej files with patch. The PURDUE >speedups caused one .rej file which I managed to get by on with some >brace format changes to the pattern it was looking for, the PURDUE+ speedups >are so full of rejected patches that I don't dare fix them all by hand. > >What happened here? Is anyone else getting nailed heavily by the PURDUE+ >speedup patch kit? If the patch files are bad, could some kind soul place >useful patch files for the PURDUE+ speedups on expo? I also had a lot of problems with one of the patch files in the Purdue speedups. Even after several messages between me and Spaf and I never managed to convince him that the patch was malformed. I also resorted to fixing the patch file to get it to work. The Purdue+ patches also generated a lot of .rej files but fortunately with them I was able to make them all work by using the -l (loose match) flag. (The compile is underway so I don't know if they really work.) -------------------------------------- rusty c. wright rusty@violet.berkeley.edu ucbvax!violet!rusty
martin@CITI.UMICH.EDU (12/15/88)
>the PURDUE+ speedups are so full of rejected patches that I don't dare fix them all by hand. No joke... sorry... >What happened here? I used "diff -c -b purdue purdue+" because I remembered reading in the xbugs-report file: > X Window System Bug Report > xbugs@expo.lcs.mit.edu > . . . > >SAMPLE FIX: >[preferred, but not necessary. Send context diffs (diff -c -b)] The -b causes blanks and tabs to be obscured beyond recognition (I guess from what happened to those patches.) >If the patch files are bad, could some kind soul place useful patch >files for the PURDUE+ speedups on expo? Yes, that has been done at about 6:30 this morning. I use good ole' (diff -c), and I even tested them! So I hope they will work for you. I got mail from a studious person saying. >I had to use "patch -l -F3" to make your patches apply... What a guy, thanks John. And, sorry for the hastle. Marty.
ekrell@hector.UUCP (Eduardo Krell) (12/15/88)
In article <14671@lll-winken.llnl.gov> brooks@maddog.llnl.gov (Eugene Brooks) writes: >Is anyone else getting nailed heavily by the PURDUE+ >speedup patch kit? Yeah. Most of the rejection are disagreements in the file ownership in the RCS id line, but some are real code differences. I decided to undo the PURDUE+ patches until someone posts the right ones. Eduardo Krell AT&T Bell Laboratories, Murray Hill, NJ UUCP: {att,decvax,ucbvax}!ulysses!ekrell Internet: ekrell@ulysses.att.com
eric@eddie.MIT.EDU (Eric Van Tassell) (12/15/88)
Does anyone know what causes the "sequence lost" messages from XlibInt.c? TIA eric
bob@allosaur.cis.ohio-state.edu (Bob Sutterfield) (12/16/88)
In article <14671@lll-winken.llnl.gov> brooks@maddog.llnl.gov (Eugene Brooks) writes: >...the PURDUE+ speedups are so full of rejected patches that I don't >dare fix them all by hand. The only rejections I got were for README, mfbbitblt.c, mfbbres.c, mfbsetsp.c, and mfbtegblt.c. They weren't really all that much trouble to hand-patch, just like the good ole days before lwall's gift to the world :-). The performance improvement is impressive. At least one of our staff has gone back to X11, even on a 4-Mb Sun-3/50. Thanks!
brooks@vette.llnl.gov (Eugene Brooks) (12/19/88)
Using the -l option to patch, I managed to get both the PURDUE and the PURDUE+ speedups to take. Although I was not "blinded" by the results, compiled with gcc 1.30, the new Xsun was notably faster.