I have now tested a fix (ignoring for now the fact that this involves an
ABI change and that one may therefore need to call the fixed package
libf2c3 instead of libf2c2; that's a can of worms I'll leave to others).
Here is what I've added to debian/rules:
elif [ $(arch) == "amd64" ] ;\
then echo "Building for AMD64" ;\
sed 's/long //' f2c.h > f2c.install.h ;\
cp f2c.install.h f2c.h ;\
That last line deserves a comment. If one omits it, then the library
ends up being built with the unpatched f2c.h, which is obviously not
the right thing to do. (It will bite you the moment you try to do any
Fortran I/O.) The corresponding stanza for the alpha architecture is
missing that line, and that looks very suspicious to me. I don't
have the hardware to test it, but I don't see why alpha and amd64
should differ in this respect.
With the fix I described above, my application (a private Debian
packaging of stsci_python,
http://www.stsci.edu/resources/software_hardware/pyraf/stsci_python )
no longer segfaults on amd64 when linked with libf2c.a.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
