On Thu, Jun 14, 2007 at 10:06:50PM +0100, Richard Jones wrote: > > Would you mind sending us (maybe as an attachment?) a list of the files > > contained in each package? From this list I can guess the content, but > Gemi did this actually. Here is the list of files: Ok, thanks. > It just lists the requirements (dependencies) of the ocaml-calendar > package. > > For comparison, here are symbols provided by the same: > > $ rpm -q --provides ocaml-calendar > ocaml(Calendar) = db8ca9a52d58e744b63d354d36f12620 Ok, so another package can depend on "ocaml(Calendar)" at the precise "version" matching the md5sum and the dep will be fulfilled only by that version of ocaml-calendar. We have on the go something similar in Debian, google for ocaml-md5sum; we will have a look at your scripts for comparison. Still we have an open issue on this: we rely on the output of ocamlobjinfo which only works on bytecode objects and hence can only ensure interface assumptions are not broken. Still, native code objects can break link time compatibility with compatible .cmis. To solve this an ocamlobjinfo-like tool would be needed for native code objects. I asked for it several time on the caml mailing list without success. IIRC Julien mentioned that such a tool is available in ocaml 3.10.0 but I haven't checked yet. How did you solve (if you did) this part of the dependency problem? Cheers. -- Stefano Zacchiroli -*- PhD in Computer Science ............... now what? zack@{cs.unibo.it,debian.org,bononia.it} -%- http://www.bononia.it/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