brian@cimage.com (Brian Kelley) (10/24/90)
I've been trying to build perl patchlevel 37 under Ultrix 4.0 without success. This is on a System 5400. The make goes fine, but 'make test' fails on tests 7 and 40. Has anyone else experienced these problems or built perl under 4.0 (I've had no problems on our Suns using the same code)? --- brian@cimage.com
grr@cbmvax.commodore.com (George Robbins) (10/24/90)
In article <1990Oct23.190637.9412@cimage.com> brian@cimage.com (Brian Kelley) writes: > > I've been trying to build perl patchlevel 37 under Ultrix 4.0 without success. > This is on a System 5400. The make goes fine, but 'make test' fails on tests > 7 and 40. Has anyone else experienced these problems or built perl under > 4.0 (I've had no problems on our Suns using the same code)? Ultrix hasn't traditionally been 100% perl clean, so having perl tests fail doesn't neccessarily reflect installation problems, but rather Ultrix bugs or perhaps legitimate (if wrong headed) differences in unix implementation. Try compiling with optimization, and see if this helps. Analyze the remaing errors to find out what the Ultrix problems are, and either patch in workarounds or file some SPR's. -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing: domain: grr@cbmvax.commodore.com Commodore, Engineering Department phone: 215-431-9349 (only by moonlite)
grr@cbmvax.commodore.com (George Robbins) (10/24/90)
In article <15359@cbmvax.commodore.com> grr@cbmvax.commodore.com (George Robbins) writes: > In article <1990Oct23.190637.9412@cimage.com> brian@cimage.com (Brian Kelley) writes: > > > > I've been trying to build perl patchlevel 37 under Ultrix 4.0 without success. > > Ultrix hasn't traditionally been 100% perl clean... > > Try compiling with optimization, and see if this helps. oops, make that "without optimization"... -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing: domain: grr@cbmvax.commodore.com Commodore, Engineering Department phone: 215-431-9349 (only by moonlite)
brian@cimage.com (Brian Kelley) (10/25/90)
In article <1990Oct23.190637.9412@cimage.com> brian@cimage.com (Brian Kelley) writes: > >I've been trying to build perl patchlevel 37 under Ultrix 4.0 without success. >This is on a System 5400. The make goes fine, but 'make test' fails on tests >7 and 40. Has anyone else experienced these problems or built perl under >4.0 (I've had no problems on our Suns using the same code)? The solution to this problem is to turn off the optimizer. Thanks to all those who responded. Brian --- brian@cimage.com
evans@decvax.dec.com (Marc Evans) (10/25/90)
In article <1990Oct23.190637.9412@cimage.com>, brian@cimage.com (Brian Kelley) writes: |> |> I've been trying to build perl patchlevel 37 under Ultrix 4.0 without success. As I have said previously, for MIPSco compilers prior to version 2.10, you will not be able to use -O. You have to use -g due to bugs in the optomizer. Ultrix 4.0 comes with version 2.0 of cc (verified by a 'cc -V foo.c | grep Mips'). The other gotcha to be sure to avoid is that configure determines that the keyword volitile is supported, when in fact the compiler chokes on it in many cases. You should edit the config.sh script when given the option at the end of Configure and 'undef' d_volitile. - Marc -- =========================================================================== Marc Evans - WB1GRH - evans@decvax.DEC.COM | Synergytics (603)635-8876 Unix and X Software Contractor | 21 Hinds Ln, Pelham, NH 03076 ===========================================================================
mcba@newt.phys.unsw.OZ.AU (Michael C. B. Ashley) (10/25/90)
In article <1990Oct23.190637.9412@cimage.com>, brian@cimage.com (Brian Kelley) writes: > > I've been trying to build perl patchlevel 37 under Ultrix 4.0 without success. > This is on a System 5400. The make goes fine, but 'make test' fails on tests > 7 and 40. Has anyone else experienced these problems or built perl under > 4.0 (I've had no problems on our Suns using the same code)? > perl patchlevel 37 passes all tests with the following modifications to the Makefile (as per the suggestions in README): (1) CFLAGS = -DLANGUAGE_C -O (2) Turn off optimization for the compilation of perly.c and tperly.c (this is easily done by editing the make rules for perly.o and tperly.o. Also, when building the Makefile with Configure, I specified volatile flags adding up to 1. I'm not sure if this is necessary, but I believe that the MIPS compiler has (or had) some trouble with volatile. Otherwise, use the default answers for everything from Configure. Good luck! Michael Ashley / Astrophysics Dept. / Uni of UNSW / mcba@usage.csd.unsw.oz.au
fuchs@it.uka.de (Harald Fuchs) (10/25/90)
grr@cbmvax.commodore.com (George Robbins) writes: >Ultrix hasn't traditionally been 100% perl clean, so having perl tests fail >doesn't neccessarily reflect installation problems, but rather Ultrix bugs >or perhaps legitimate (if wrong headed) differences in unix implementation. >Try compiling with optimization, and see if this helps. Nah. There are two problems with the Ultrix C compilers: 1. They aren't quite ANSI. Especially, they have problems with 'void'. Perl Configure guesses wrong about that. Answer '1' to the 'void' question. 2. The optimizers are buggy. Answer 'none' when Configure asks for the C compilation flags. That should do it. -- Harald Fuchs <fuchs@it.uka.de> <fuchs%it.uka.de@relay.cs.net> ... <fuchs@telematik.informatik.uni-karlsruhe.dbp.de> *gulp*