[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: autoconf 2.52
From: |
Akim Demaille |
Subject: |
Re: autoconf 2.52 |
Date: |
09 Aug 2001 10:03:46 +0200 |
User-agent: |
Gnus/5.0808 (Gnus v5.8.8) XEmacs/21.4 (Academic Rigor) |
>>>>> "Steven" == Steven G Johnson <address@hidden> writes:
Steven> The problem here seems to be that your f77 script does not
Steven> support any known "verbose" flag to get it to say what
Steven> libraries it is linking with, so autoconf cannot detect them.
Steven> (There is usually a libf2c or something that is required if
Steven> you want to link to any code using Fortran intrinsic
Steven> functions, etcetera.) We have two options here:
Steven> 1) If there is some special verbose flag that your f77 script
Steven> uses, we can easily support it. Currently, we check: -v
Steven> -verbose --verbose -V -\#\#\#
Steven> 2) We can try to manually detect this f77-f2c script and
Steven> manually set the right FLIBS (Fortran libraries). Ugh.
Steven> 3) We can screw it and say that linking C and Fortran code
Steven> using this script is not supported automatically. The user
Steven> has to specify the FLIBS environment variable explicitly to
Steven> get the right Fortran libs.
But what should be the policy wrt failure? Does it make sense to keep
just a warning? Shouldn't it be an error?
In fact, since it means this is a possible failure, I'm just willing
to know _how_ it can fail so that I can adjust the test suite to
correctly detect this is a possible failure.