Am Mittwoch 10 November 2004 18:02 schrieb Michael Nottebrock:
> On Tuesday, 9. November 2004 18:17, Szombathelyi György wrote:
> > I hope this small(?) inconvenience will not end in removing the
> > --as-needed patch (if you're a developer, and disturbed by this behavior
> > you can always compile with --disable-as-needed),
>
> I agree it is useful, but like closure and final, it's not obviously not
> totally guaranteed-to-work-no-matter-what - why not handle it like closure
> and final and make --disable the default and make things consistent. There
> will be always enough developers and packagers enabling it and thus
> preventing it from rotting.
closures were enabled for a very long time and were disabled as enough
systems support -Wno-undefined to make it not necessary any more (beside
for template support on some compilers).

And final is just no toy, as it takes much too much memory for a background
compile job.

The situation with as-needed is a bit different: it can actually break 
something, but we're not sure yet what. As far as I understood Lauri
the nvidia problem was there before, it only triggered with the linker flag.
And unless we know what the problems are I see no point in hiding the flag.
If it appears to cause unmaintainable trouble, we can still disable it, but I 
don't see that yet.

Greetings, Stephan