| 1 |
Il giorno mar, 05/10/2010 alle 13.04 +0200, Angelo Arrifano ha scritto: |
| 2 |
> There are a lot of packages that need this information to correctly link |
| 3 |
> against libtool managed libraries, for example, there are packages that |
| 4 |
> linked against GL but didn't set -lGL -lGLU because it was relying on |
| 5 |
> libtool to get that information (guess from where?). |
| 6 |
|
| 7 |
Definitely not from /usr/lib/libGL.la given that libGL is part of mesa |
| 8 |
and mesa did not use libtool for linking until very recently, so it's |
| 9 |
either something relying on just the most recent versions of mesa, or |
| 10 |
it's screwed. For your information, mesa has supported pkg-config for a |
| 11 |
longer time than it used libtool to build. |
| 12 |
|
| 13 |
Other interesting note: libGL on non-Xorg systems has no reason to use |
| 14 |
libtool at all, even today. |
| 15 |
|
| 16 |
Where does the linker find this information? For dynamic linking, from |
| 17 |
the ELF files. The problem sticks for static linking, but I have my |
| 18 |
sincere doubts that anyone in his sane mind is going to statically link |
| 19 |
libGL for the simple reason that it can depend on the driver used (mesa, |
| 20 |
nvidia, ATI, ...) and might even use dynamic linking against other |
| 21 |
backends. |
| 22 |
|
| 23 |
> Mind you that the community is wider than one can imagine. I happen to |
| 24 |
> work in the academia and I know a lot of nasty stuff people do to save |
| 25 |
> time (at least is what they think) for deadlines. As a user, I would |
| 26 |
> hate to have my research program/script broken just because some dev |
| 27 |
> decided to make the distribution I use his personal sandbox. |
| 28 |
|
| 29 |
No, you'd just have to do things sanely. I sincerely don't see Gentoo |
| 30 |
needing to not move forward (or add much more complexity in the already |
| 31 |
complex mix) just because some student decided that it's cooler to hack |
| 32 |
something up rather than learning how the thing is done to begin with. |
| 33 |
|
| 34 |
We already don't support many other things that do break hacky scripts |
| 35 |
that are written in academic (and not just) circles; does that mean we |
| 36 |
have to revert those and pad around everything for our users, lest some |
| 37 |
thing that never should have worked actually stop working? |
| 38 |
|
| 39 |
> Besides, doesn't this kind of changes belong in upstream and then |
| 40 |
> eventually come to the distros? Why don't you make a patch and send |
| 41 |
> upstream if these libtool files are so useless? |
| 42 |
|
| 43 |
I see you haven't read my post on the matter I linked at the root of the |
| 44 |
thread. Please do so and don't ask me questions I answered already |
| 45 |
there. |
| 46 |
|
| 47 |
http://blog.flameeyes.eu/2010/10/04/libtool-archives-and-their-pointless-points |
| 48 |
|
| 49 |
-- |
| 50 |
Diego Elio Pettenò — “Flameeyes” |
| 51 |
http://blog.flameeyes.eu/ |
| 52 |
|
| 53 |
If you found a .asc file in this mail and know not what it is, |
| 54 |
it's a GnuPG digital signature: http://www.gnupg.org/ |
Archives