[petsc-dev] problems with shared libraries and pragmatic.py

[Jed Brown]

Satish Balay <balay at mcs.anl.gov> writes:

> On Thu, 15 Oct 2015, Jed Brown wrote:
>
>> Matthew Knepley <knepley at gmail.com> writes:
>
>> But is this the only way?  Why on earth should the person distributing a
>> shared library be the one that needs to make a decision about whether a
>> *user* of that library links using RPATH or not?
>
> I believe thats an OSX requirement. I think its more like - the
> library should be able to find itself :)

But you have to find it at link time anyway.  At that time, I (the user)
has found it.  Why can't I say I want to use that version?

This sounded like deja vu, so I checked the archives and found this
link:

https://www.mikeash.com/pyblog/friday-qa-2009-11-06-linking-and-install-names.html

Why don't all libraries set their install name to @rpath/libfoo.dylib so
that the application linking to them can set -rpath appropriately?  Why
would anyone want this crap hard-coded in the library?  It seems
backwards.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 818 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20151015/068c3608/attachment.sig>