| 1 |
On 13-06-2007 12:08:49 +0200, Michael Haubenwallner wrote: |
| 2 |
> On Tue, 2007-06-12 at 11:20 +0200, Fabian Groffen wrote: |
| 3 |
> > 1) the patch was reverted, because there is no CVS header in the patch, |
| 4 |
> > so my update script can't do anything but making it the same as the main |
| 5 |
> > tree version, in this case overwriting the differences |
| 6 |
> > 2) I tried applying the patch, but it failed to apply somehow |
| 7 |
> > 3) the patch adds some GNU binutils specific linker stuff AFAICT, so |
| 8 |
> > maybe it is better to apply it for the archs we use the GNU linker |
| 9 |
> > with instead of not applying it for the archs we don't use it? |
| 10 |
> |
| 11 |
> Agreed, but |
| 12 |
> |
| 13 |
> 4) Some native linkers (hpux fex.) do have soname support, but with |
| 14 |
> different argument (+h for hpux), not '-soname'. |
| 15 |
> And we cannot use GNU ld on hpux. |
| 16 |
> |
| 17 |
> Maybe ldwrapper should map "-soname=name" to "+h name" for hpux once, |
| 18 |
> and for aix/darwin, just drop "-soname=name" ? |
| 19 |
|
| 20 |
How tricky would it be to do this? Can it go wrong? With the modular |
| 21 |
setup of the ldwrapper it is completely instance specific of course... |
| 22 |
How much other things can we "solve" this way? hmmm.... |
| 23 |
|
| 24 |
|
| 25 |
-- |
| 26 |
Fabian Groffen |
| 27 |
Gentoo on a different level |
| 28 |
|
| 29 |
-- |
| 30 |
gentoo-alt@g.o mailing list |
Archives