From help-octave-request at bevo dot che dot wisc dot edu  Wed Oct  7 12:07:38 1998
Subject: Re: Compilation problems under Solaris 2.6
From: todd at unagi dot es dot hac dot com (Todd Ellenberger)
To: jwe at bevo dot che dot wisc dot edu, help-octave@bevo.che.wisc.edu
Date: Wed, 7 Oct 98 10:08:35 PDT

Thanks for all your help, John. Today I sucessfully compiled and installed octave,
and it looks great! The best tip came on todays postings to the info list, namely
the message from veco at montefiore dot ulg dot ac dot be Re: Octave 2.1.7 on solaris 2.6 x86
saying that gnu make 3.77 does not result in the fortran libraries being compiled
when building octave. I checked the output from my previous failed attempt, and sure
enough the g77 compiler was never called. I installed make version 3.75 and sure enough
during the build all the fortran libraries in libcruft were compiled and built into
libcruft.so. In the previous (failed) attempts, libcruft.so was created, but it must
have been void of any code. I probably could have manually compiled the libraries into
.o files, and then run the big octave make process under make 3.77.

Thanks for all your help - I look forward to using and promoting octave at my company
for much of the analytical work we normally handle with Matlab.

Todd Ellenberger
(todd at unagi dot es dot hac dot com)

I will also post this to the help mailing list, as it may save someone else some time.


|| It appears that g77 is appending the underscore. Is there any way to tell if
|| these functions are being properly built into libcruft.a and libcruft.so?

| Try
|
|  nm libcruft.a
|
| or
|
|  nm libcruft.so
|
| If the symbols seem to be there, can you send me the final link
| command again?
|
| Is it possible that there is another older copy of libcruft installed
| somewhere on the system and that the linker is choosing the wrong one?
|
| jwe