On Sun, May 04, 2008 at 09:32:16PM +0000, Sylvain Le Gall wrote:
> What about patching bin-prot syntax extension to produce "nothing" when
> used... This way, everything still get compiled with "fake" bin-prot...
> I think it will make the required patch smaller (i.e. you will have only
> to patch where the serialization is used and not everywhere).
Nice idea. Actually it will change what is being patched, as we will
then patch bin-prot rather than core, but this sounds like a good idea.
Still, it won't be enough to remove completely the need to patch core,
as core has module interfaces which assume that bin-prot has generated
something. One can then think to push your approach further and make
bin-prot generate "assert false" functions rather than nothing, having
static type correctness but runtime errors (which then would become
harder to detect by buildds). Well, it seems like an interesting path,
though not entirely trivial.
Still, the best would be for bin-prot to support all archs. Maybe we
should wait for the next round of upstream releases? I do hope that
various people will push and/or push for supporting more archs. If
nothing will change, we can move from the current very hackish solution
to yours less hackish solution? (provided we can implement it properly)
Cheers.
--
Stefano Zacchiroli -*- PhD in Computer Science ............... now what?
zack@{upsilon.cc,cs.unibo.it,debian.org} -<%>- http://upsilon.cc/zack/
(15:56:48) Zack: e la demo dema ? /\ All one has to do is hit the
(15:57:15) Bac: no, la demo scema \/ right keys at the right time
Attachment:
signature.asc
Description: Digital signature