On Mon, Jun 11, 2007 at 12:42:47PM -0400, Eric Cooper wrote:
> Previously this package produced only libsyslog-ocaml-dev, since it
> was pure OCaml. Now that it's impure, do I need to produce
> libsyslog-ocaml also? I can't remember in what situations it's
> required, and a quick test of a bytecode client of syslog seemed to
> work fine. The packaging policy seems to say that it's needed, but
> not *why* it's needed.
It is required if you want to be able to distribute a portable bytecode
executable (Arch: all) which uses the C stub directly from
libsyslog-ocaml (which will be Arch: any). You can avoid it by
distributing bytecode programs linked in the so called custom mode, but
such executables are not portable, but rather architecture specific as
they contain an embedded OCaml bytecode interpreter.
The -ocaml library is pointless for native code no matter what.
So the policy states to create them to enable users to create portable
bytecode programs against your library.
> My second question is somewhat higher level. The bugfix above is
> Linux-specific, so upstream doesn't want to adopt it (if he moves to
> some kind of autoconfiguration tool in the future he might). Instead,
> he'll use the first solution that I came up with, which is pure OCaml
> but less efficient (installing and removing a Unix signal handler
> around each write call).
>
> Do you think the more efficient version is worth having a
> Debian-specific patch and (possibly) a 2nd package in the
> archive?
I think it is worth, but this is of course just my 0.02€.
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