| 1 |
On Fri, 30 May 2008 18:29:49 +0200 |
| 2 |
flameeyes@×××××.com (Diego 'Flameeyes' Pettenò) wrote: |
| 3 |
> David Leverton <levertond@××××××××××.com> writes: |
| 4 |
> > It's not just the order, it also breaks things that rely on |
| 5 |
> > the .init section being called at all to register themselves with |
| 6 |
> > the core application (with --as-needed, the .so doesn't get loaded |
| 7 |
> > in the first place, so it doesn't get a chance to run anything). |
| 8 |
> |
| 9 |
> This really is backward, solution-wise: you expect the "core |
| 10 |
> application" to know enough of the plugins to link them together, but |
| 11 |
> not enough to call their init functions... |
| 12 |
|
| 13 |
Actually, no. The core application doesn't have to be doing the |
| 14 |
linking. The linking can be done by an intermediate library. |
| 15 |
|
| 16 |
> And still, it breaks not only with --as-needed but also with the Sun |
| 17 |
> linker, with the PE file format and other non-ELF file formats. So |
| 18 |
> it's really not a problem _limited_ to --as-needed. |
| 19 |
|
| 20 |
It works with any standard-compliant C++ setup. |
| 21 |
|
| 22 |
> Beside, you can _force_ a link by using --no-as-needed before a given |
| 23 |
> library, that's what it's supposed to happen if you don't want to be |
| 24 |
> helped by the linker. |
| 25 |
|
| 26 |
And next you'll be saying "and you can force a compile using |
| 27 |
--no-broken-behaviour-that-goes-against-the-standard". |
| 28 |
|
| 29 |
> To a very minimum this can be said to be a _clash_ between two |
| 30 |
> designs. Saying that --as-needed is _broken_ because it breaks this |
| 31 |
> case is _quite_ an exaggeration... |
| 32 |
|
| 33 |
Saying it's broken because it goes against an international standard |
| 34 |
isn't... |
| 35 |
|
| 36 |
-- |
| 37 |
Ciaran McCreesh |
Archives