Re: migration of OCaml 3.10 into testing
On Mon, 2007-11-05 at 12:05 +0100, Stefano Zacchiroli wrote:
> On Mon, Nov 05, 2007 at 10:12:03AM +0000, Samuel Mimram wrote:
> > Now that the new gtk is in testing, we should try to have OCaml 3.10 in
> > testing. From [1], it seems that we only need to request a removal of
> > felix. Do you see any other blocker?
>
> Nothing else apparently.
>
> Who can take care of asking/checking the needed hints?
>
> <felix_rant>
> Mike: I see that recently you've re-uploaded felix, can I ask why you
> didn't tackle the build issue on some architectures? At the very minimum
> we can avoid building the package on that arch, but also falling back to
> bytecode only there shouldn't be too hard. Am I missing something else
> here?
> </felix_rant>
>
> As I've already said elsewhere, I totally agree with removing felix from
> testing.
Please don't remove Felix when the problem is known to be
Ocaml native code compiler on those arches. Felix automatically
uses that if it is available.
Removing Felix just hides the problem that Ocaml itself
is broken. You should remove Ocaml native code compiler instead.
If you remove Felix from testing entirely, the latest version
won't go out in a timely manner on platforms where Ocaml native
code compiler does work correctly, particularly x86 and x86_64.
Felix causes problems for the simple reason it actually
runs regression tests as part of the build. If people
actually tested the code built, I suspect quite a few
more packages would be found to be broken.
--
John Skaller <skaller at users dot sf dot net>
Felix, successor to C++: http://felix.sf.net
Reply to: