dpm@cs.cmu.edu (David Maynard) (03/21/90)
Has anyone generated a list of R4 client (and demo/example) programs that need to be compiled with the -fwritable-strings option of GCC. So far I've identified xterm and puzzle (which would occasionally dump core), but the other clients I've tested appear to work fine. I read in the (outdated) X11.README that came with GCC that twm might need writable strings, but it "seems" to work without them. I'm working with GCC 1.37.1 on a Sun-2 under SunOS 3.5 with MIT fixes 1-6. On the same subject...what is the "best" way to specify special compiler options for a particular client? My quick and dirty way was to specify an explicit "CC = gcc..." line in the appropriate Imakefiles. That obviously isn't the best way to specify things since it won't automatically pick up global changes I make to the compiler options. P.S. I'm impressed! The R4 "make World" only took ~19 hours on a 4 meg Sun-2/120. If I recall correctly, the R3 compile took closer to 25 hours. --- David P. Maynard (dpm@cs.cmu.edu) Dept. of Electrical and Computer Engineering Carnegie Mellon University, Pittsburgh, PA 15213 ---
rmtodd@uokmax.uucp (Richard Michael Todd) (03/21/90)
In article <1990Mar20.191933.1792@cs.cmu.edu> dpm@cs.cmu.edu (David Maynard) writes: >Has anyone generated a list of R4 client (and demo/example) programs >that need to be compiled with the -fwritable-strings option of GCC. So >far I've identified xterm and puzzle (which would occasionally dump >core), but the other clients I've tested appear to work fine. I read in I found that xfd also needed it. This is on a Mac IIx under A/UX 1.1, GCC 1.37, MIT X11R4 + fixes 1-4. Haven't seen any problems with twm. puzzle would *always* dump core without -fwritable-strings, xterm would always do so when switching to Tek mode. >P.S. I'm impressed! The R4 "make World" only took ~19 hours on a 4 meg >Sun-2/120. If I recall correctly, the R3 compile took closer to 25 hours. It took ~12 hours on my Mac IIx with 4M of memory. From the system stats programs I ran it looks like a good deal of the time was spent paging (gcc is a big memory hog). Haven't tried a Make World since I upgraded to 8M, but I suspect it'll go a whole lot faster... -- Richard Todd rmtodd@chinet.chi.il.us or rmtodd@uokmax.ecn.uoknor.edu
rcb@ccpv1.ncsu.edu (Randy Buckland) (03/21/90)
rmtodd@uokmax.uucp (Richard Michael Todd) writes: >In article <1990Mar20.191933.1792@cs.cmu.edu> dpm@cs.cmu.edu (David Maynard) writes: >>P.S. I'm impressed! The R4 "make World" only took ~19 hours on a 4 meg >>Sun-2/120. If I recall correctly, the R3 compile took closer to 25 hours. > It took ~12 hours on my Mac IIx with 4M of memory. From the system stats >programs I ran it looks like a good deal of the time was spent paging (gcc >is a big memory hog). Haven't tried a Make World since I upgraded to 8M, >but I suspect it'll go a whole lot faster... It's statistics/benchmark time. What was the compile times for tape1 of X11R4 on your machine? For a Vaxstation 3100, 16M memory, Ultrix 3.1, Ultrix cc compiler, time was just over 5 hours. How about the times on some of the RISC machines? This seems like a good real world test of performance. Randy Buckland rcb@ncsuvx.ncsu.edu