casey@lll-crg.llnl.gov (Casey Leedom) (12/30/88)
Well I've perused the news group and there's no other mention of this, so ... I finally bit the bullet and installed the Purdue and PurduePlus speed ups. The reviews of the performance enhancements were just too attractive. Not even the warning that the PurduePlus speed ups would break my ability to compile X11.3 with anything but gcc deterred me (apparently the #else halves of the PurduePlus #ifdef __GNUCC__ code is not correct) ... And, I have to say that the reviews were right. It's wonderfully fast. I recommend them to anyone willing to be committed to compiling X with gcc and not being able to compile with your native compiler. Just be sure that you don't blithely install a new version of gcc without checking to make sure that it compiles X without bugs ... But, on to the main point, everything except xclock -analog seems to be working. Not understanding the essential nature of the changes to the server, I'm totally in the dark and unable to find the problem myself. In analog mode, all line segment seem to be rooted to the left hand edge of the window giving a ``Death Star'' like appearance. Very pretty and surrealistic, but not really useful. Salient details: Purdue, PurduePlus, and MIT patches 1, 2, and 3 installed; compiling with GNU cc 1.32. Sun 3/280, 2/50, 3/60. Sun OS 3.5. Casey
casey@gauss.llnl.gov (Casey Leedom) (12/30/88)
| From: casey@lll-crg.llnl.gov (Casey Leedom) | | And, I have to say that the reviews were right. The Speed ups make X | wonderfully fast. I recommend them to anyone willing to be committed to | compiling X with gcc and not being able to compile with your native | compiler. Just be sure that you don't blithely install a new version of | gcc without checking to make sure that it compiles X without bugs ... | | But, on to the main point, everything except xclock -analog seems to be | working. Well, cut me a new point of view if I can't follow my own advise. I installed the server speed ups at the same time I upgraded from gcc 1.31 to 1.32. Gcc 1.32 apparently has a bug in it which causes the behavior in xclock -analog that I mentioned. I will endeavor to find the bug and report it to the GNU people. Sorry for wasting everyone's time ... Casey