On Mon, Jul 08, 2013 at 10:51:10AM +0100, Gianfranco Costamagna wrote:
> > Yes, but what does that have to do with the commit you linked? Doesn't
> > it just add the symlink?
> 
> I honestly don't know, linking it locally makes gcc (4.6) to fail the linking 
> of the file.
> without the link gcc automatically finds the right location without any 
> problems 

Well, maybe I'm missing something. Let me say what it looks like:

* The BOINC app samples link against libstdc++ statically on x86 because
  they're meant to run on the *clients*, and Linux clients might not have
  the right libstdc++ available. The binaries are not meant to run on the
  server machine!

* The commit you link fixes the build, by adding a symlink to libstdc++.a,
  but it doesn't add libstdc++.a to the $(CXX) command line: it's already
  there (and it's been there for a long time, for the reason above).

* Your patch removes libstdc++.a entirely, from the $(CXX) command line,
  which means you get binaries dynamically linked against libstdc++.
  That doesn't revert the commit you linked, but instead makes a
  different change (one which produces binaries which might not work
  on old x86 client Linux machines).

If that last bit is wrong then, sorry, I've misinterpreted your patch.

But it does seem also that building the app samples in the Debian/Ubuntu
packages is pointless, since you want to build them targetting an old
libc anyway, right? So usually they should be built on a VM or in a
chroot. So Bernd Machenschalk's suggestion for a --disable-example-apps
or something might be a good idea, I don't know.

- Alyssa 
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.

Reply via email to