Re: (1) fr.po trivial; (2) automatic URLs for plain text format?

[Dmitry Borodaenko]

On 11/30/06, boud <address@hidden> wrote:
> You're right: /usr/lib/ruby/1.8/uri/ has the necessary stuff.
>
> How does this look:  patch-autoURL-0.2 ?
>
> Any chance of this getting into 0.6.0? ;)

Err, 0.6.0 is out already. I've sent announcements to all possible
places, but only sent it to samizdat-devel and imc-{cms,tech} today.
Sorry about that, it's been a while since the previous release, so I
kind of forgot how to do it properly :)

Take a look at the auto-url code I've put into the release, I found
out that it should be a bit more complicated: we need to limit it to a
certain list of URL schemes so as to avoid false positives, and to
avoid grabbing punctuation that goes after the URL, because people are
much more likely to write something like "look http://here/, and there
(ftp://too/)", than to really mean URLs to end with punctuation
characters.

> And what time scale are you thinking of for 0.6.0?  i might try a
> spanish translation, since that would particularly be useful for the
> indymedia community...

Please do, even though release is out (and I do hope we're not too
late to get it into Debian/etch), I plan to continue uploading regular
snapshots into experimental.

> BTW, i think that somewhere at the beginning of the uri_list.each loop
> will in the future be a convenient place to call a future routine
> which will trap people putting in wiki/blog link spam - but that will
> also need to be used for the textile and html formats.

IMHO the right place for this, as well as other abuse checks, is
PublishMessage class which already does a lot of checks on published
messages. I would simply pass the whole content of any inline message
through a bayesian filter.

--
Dmitry Borodaenko