upstream update nitify [Was: maybe-ITP: bsdiff]

[John Morrison]

On Fri, January 27, 2006 11:16 pm, Lapo Luchini wrote:
> John Morrison wrote:
>> Freshmeat do a number of RSS feeds see <http://freshmeat.net/backend/>
>> for
>> the list.
>>
> Mhh, neat.
> Problem is: freshmeat is not /always/ updated.

How about <http://distrowatch.com/news/dwp.xml>? or scraping
<http://distrowatch.com/packages.php>?

> rsync, as an example, is at release 2.6.0, while truly it is a 2.6.6
> last time I (manually) checked.

Distrowatch reports 2.6.6

>> Perhaps <http://rss.freshmeat.net/freshmeat/feeds/fm-releases-unix>
>> would
>> be the best?  Can perl consume RSS feeds?
>>
> IMHO the best is to stick with the URL+regular expression form: it an
> RSS is available it's only easier to "parse" (less thing on the "page"
> that may falsely match the r.e.)
>
> While using the official homepage:
> http://rsync.samba.org/
>
> % links -source "http://rsync.samba.org/" | egrep -o
> "[0-9]+\.[0-9]+\.[0-9]+ released" | head -n 1
> 2.6.6 released
>
> Parsing the FTP directory would be even less error-prone, if available.
>
> Uhm... an approach like this can't as reliable as being "the final anwer
> to update notifications", but IMHO can surely help.
> More so if each package has /more/ than one address (frashmeat,
> homepage, ftp) so that if some of the sources fail or change format, the
> others are still up (a different page could show red lines in the ones
> that failed to parse the last few days, something like that).
>
> My approach to that would probably include PHP and a sqlite database.
> If there is some interest in it I could develop and mantain it.

Fair enough - I never (well, rarely!) argue with somebody who's willing to
put the effort in :)

J.