Re: libdebctrl Update
Russ:
On Tue, Jun 30, 2009 at 12:16 PM, Russ Allbery<rra@debian.org> wrote:
> Jonathan Yu <jonathan.i.yu@gmail.com> writes:
>> On Tue, Jun 30, 2009 at 2:24 AM, Russ Allbery <rra@debian.org> wrote:
>>> Jonathan Yu <jonathan.i.yu@gmail.com> writes:
>
>>>> c) The library could clean up such strange things before writing the file
>
>>> Yeah, although in some cases the meaning is ambiguous. I'm not sure
>>> if you want to just strip out ambiguous or bad data. But duplicates
>>> it could definitely remove.
>
>> Where are meanings ambiguous?
>
> I was thinking of things like unsatisfiable conditions or unknown
> architectures, where it's hard to be sure what the user really meant.
Yeah, that's a really good point. It's currently just a warning --
just a flag to sort of say "I don't know of that architecture" -- but
to clarify, it will still accept the architecture. It'll just be a bit
confused until libdebctrl is upgraded :-)
As you mentioned, people don't need to announce new architectures so
there are cases where even dpkg-architecture wouldn't know about a
given architecture, but nonetheless it should be valid. The warnings
might get a bit annoying, but I think they do more good than harm on
the whole.
The idea is to future-proof libdebctrl and accept anything the user
gives us, but also to let them know if something looks weird. I want
to always make sure that if you just read in a file and then print it
back out, they are /semantically/ equivalent, even if the semantics
have been changed a bit.
Thanks again :-)
Hope this explains stuff for everyone! I'll be sending an update to
the debian-perl list again shortly, too.
Cheers,
Jonathan
Reply to: