An experience syndicating Trac Timeline RSS feeds with PlanetPlanet

[Sam Ruby]

Alvaro del Castillo wrote:
> Hi guys!
> 
> First of all thanks for your work in PlanetPlanet. I have used it
> several times without problems ... but I have found some interesting
> tricks trying to use it with Trac Timeline RSS feed:
> 
> http://qualoss.libresoft.es/cgi-bin/trac.cgi/timeline
> 
> I have integrated several projects timeline feeds in one page using
> PlanetPlanet:
> 
> http://projects.libresoft.es/
> 
> The problem has been that PlanetPlanet in order to create a unique
> identifier and store the entry feed in the cache, it uses:
> 
> ---------------------------
> for entry in entries:
>             # Try really hard to find some kind of unique identifier
>             if entry.has_key("id"):
>                 entry_id = cache.utf8(entry.id)
>             elif entry.has_key("link"):
>                 entry_id = cache.utf8(entry.link)
>             elif entry.has_key("title"):
>                 entry_id = (self.url + "/"
>                             + md5.new(cache.utf8(entry.title)).hexdigest())
>             elif entry.has_key("summary"):
>                 entry_id = (self.url + "/"
>                             + md5.new(cache.utf8(entry.summary)).hexdigest())
>             else:
>                 log.error("Unable to find or generate id, entry ignored")
>                 continue
> ---------------------------
> 
> In the RSS feed from Trac timeline several entries share the same link
> (for example, to go to a modified Wiki pages) so using this identifier
> makes PlanetPlanet cache to overwrite different RSS entries and broke
> the Trac integration we are looking for.
> 
> In order to solve the problem we have added to the entry link field, the
> date field also, so all the entries are different for PlanetPlanet and
> Trac RSS is shown completely.
> 
> ------------------------------
> +++ __init__.py 2006-10-15 10:22:20.266702153 +0200
> @@ -739,7 +740,7 @@
>              if entry.has_key("id"):
>                  entry_id = cache.utf8(entry.id)
>              elif entry.has_key("link"):
> -                entry_id = cache.utf8(entry.link)
> +                entry_id = cache.utf8(entry.link + entry.date)
>              elif entry.has_key("title"):
>                  entry_id = (self.url + "/"
>                              + md5.new(cache.utf8(entry.title)).hexdigest())
> ------------------------------
> 
> I understand that PlanetPlanet has to fight with RSS feeds with
> duplicates entries and this filter is good for other feeds but maybe you
> should try to enable this option in someway in the PlanetPlanet config
> so the good RSS feeds can pass trough PlanetPlanet cache without
> problems.
> 
> I am not subscribed to the list so for comments about this thread please
> CC me.
> 
> Cheers and thanks for your work guys!

This patch would cause significant problems on feeds without item level 
dates, for example:

     http://partners.userland.com/people/docSearls.xml

Yes, yet another config option looks like the way to go.

- Sam Ruby