| 1 |
On Sat, 31 May 2008 03:03:42 +0200 |
| 2 |
Luca Barbato <lu_zero@g.o> wrote: |
| 3 |
> Ciaran McCreesh wrote: |
| 4 |
> > Which is where the design flaw is -- as-needed incorrectly assumes |
| 5 |
> > that the only type of dependency between shared objects is a name |
| 6 |
> > dependency. This isn't true with C++ static initialisers. |
| 7 |
> |
| 8 |
> I don't see why should be different than abusing .init in any other |
| 9 |
> language that let you do (ok, C, C++, asm mostly). |
| 10 |
|
| 11 |
In C++ it's not abuse. It's using the language as specified and |
| 12 |
designed. |
| 13 |
|
| 14 |
> > Unfortunately, the ricers shoving as-needed upon everyone aren't |
| 15 |
> > smart |
| 16 |
> |
| 17 |
> Asking people to not do stuff that is unportable (Solaris and PE |
| 18 |
> based systems) seems sensible and not ricing. |
| 19 |
|
| 20 |
Not where "unportable" means "doesn't work on systems that fail to |
| 21 |
correctly implement widely used international standards" it doesn't. |
| 22 |
|
| 23 |
You might as well say "you shouldn't use lchown() because BSD 2 doesn't |
| 24 |
support it". |
| 25 |
|
| 26 |
> > enough to fix libtool, which is the real problem here, so they go |
| 27 |
> > for the thing they think they understand instead, without thinking |
| 28 |
> > the implications through -- as-needed, like fast-math, is for |
| 29 |
> > programs explicitly designed for it, not for universal use. |
| 30 |
> |
| 31 |
> Differently -ffast-math is setting up a slightly different behavior |
| 32 |
> than the usual standard, --as-needed enforce what is the default |
| 33 |
> standard in determined architectures, thus the exception and the |
| 34 |
> universality are quite reverted. |
| 35 |
|
| 36 |
Both are standard-violating options that are useful for applications |
| 37 |
designed to work with them explicitly. |
| 38 |
|
| 39 |
> We already started to think how to fix libtool, or at least make it |
| 40 |
> less annoying, removing .la files when they are not necessary. |
| 41 |
|
| 42 |
Again, that's silly ricing. Saving a few kBytes is irrelevant. Instead, |
| 43 |
you should be focusing your efforts upon something that will really |
| 44 |
make a difference, like getting something based upon this into upstream: |
| 45 |
|
| 46 |
http://patches.ubuntu.com/by-release/extracted/debian/libt/libtool/1.9+20051221-1/link_all_deplibs.dpatch |
| 47 |
|
| 48 |
-- |
| 49 |
Ciaran McCreesh |
Archives