Joe:
Well, perhaps there was a problem because you elected to use a different directory to the default (ie. /usr/local)? OTOH, I would expect that one of the steps should have stopped with an error message, if there was a problem?
Thanks. No errors to indicate that. OTOH there is so much stuff going by on the screen - how is one to know if a warning is a problem?
I get
../../gcc-2.95.2/gcc/toplev.c:1178: warning: initialization from incompatible po is that a problem?
I was able to compile without problems by following the instructions verbatim (except that I chose "/temp/build" rather than "/home/fred" as the working directory and GCC 2.95.3 instead of 2.95.1). So, it does work and the instructions are correct. I have Mandrake 8.0 and build GPC based on the SuSE GCC source code (I was logged as "root", so I didn't need to worry about write privileges).
BTW, shouldn't "gpc-20020410" be "gpc-20020510", instead?
I wish. I modified the code of gpcbuild so it would not make a mistake like that again. The problem was that, unfortunately, the directory inside package.tar.gz is NOT 'package'; that was a pain until I realized I could use tar tzf to get the current name. (The intent of gpcbuild is that it can just figure out all the names by itself. If there is an upgrade, I'll just run it again and get a complete build next time with one command. Only ... it doesn't work yet! :-)
So it now seems to run just fine, doing all the steps. There is a cryptic statement at
http://www.gnu-pascal.de/gpc_51.html#SEC51
to 'Make sure that GNU make is installed.' A clue as to how to check that would be useful!!! Please add something there in the manual, thanks.
To be safe I switched to gmake but it made no difference.
Does my script work on other systems? (It gives lots of errors on the c compile on solaris ... don't these scripts detect their environment?)
Tom
Dr. Thomas D. Schneider National Cancer Institute Laboratory of Experimental and Computational Biology Frederick, Maryland 21702-1201 toms@ncifcrf.gov permanent email: toms@alum.mit.edu http://www.lecb.ncifcrf.gov/~toms/
toms@ncifcrf.gov wrote:
Well, perhaps there was a problem because you elected to use a different directory to the default (ie. /usr/local)? OTOH, I would expect that one of the steps should have stopped with an error message, if there was a problem?
Generally, the build and install process should work with any target directory that you have write access to. If the directory is FOO, you'll find the main binaries (gpc, etc.) in FOO/bin. Some other files needed by it are in FOO/lib/gcc-lib/... The exact location is hard-coded in the gpc executable, so it won't work (without setting some extra variables) if you move them afterwards.
Thanks. No errors to indicate that. OTOH there is so much stuff going by on the screen - how is one to know if a warning is a problem?
Try `make -s' for a start. The remaining messages are either produced by the GCC frontend (whose developers don't seem too interested in getting rid of them; you might want to complain to them), or real GPC problems.
I get
../../gcc-2.95.2/gcc/toplev.c:1178: warning: initialization from incompatible po is that a problem?
No, that's one of the first category (the most common one).
So it now seems to run just fine, doing all the steps. There is a cryptic statement at
http://www.gnu-pascal.de/gpc_51.html#SEC51
to 'Make sure that GNU make is installed.' A clue as to how to check that would be useful!!!
make --version
If it says it's GNU make, then it is. Otherwise (it says something else or complains about the unknown arguments), it's not.
To be safe I switched to gmake but it made no difference.
Yes, gmake (where it exists) is usually GNU make. Did you start from scratch using it? Otherwise, if something made by another make is left, this might cause problems.
(It gives lots of errors on the c compile on solaris ... don't these scripts detect their environment?)
What scripts, what errors? Your statement is rather vague ...
I'm not sure if anyone tried to build GPC with a non-gcc compiler recently (if so, please give me a note), so there might in fact be some errors. If I know what they are, I might be able to fix them.
Frank
-
Frank Heckenbach -
toms＠ncifcrf.gov