Re: [Pyblosxom-devel] Egads! Bugs?

['Wari W. <wa...@ce...>]

* will <wi...@bl...> [030205 13:16]:
> I've had lots of experience with SourceForge over the last 3 years and
> I really hate their documentation manager and forums system.  Tracking
> bugs and features sucks unless you have a really small amount of them.
> Now granted it's free so it's lame to complain much.

Any idea how to remove some of those services, like forums and
documentation manager from my sourceforge page?

> I once wasn't a fan of the wiki, but after playing with it this last
> weekend I agree with Wari that it's a decent system for our
> documentation.  It just needs some more content.  :)

I know the documentation isn't great, and I'm extremely jealous of
Rael's documentation :) But I can't just copy and paste from his site :)
Most of his documentation do apply for pyblosxom as well.

> In terms of the url handler, I can jerry rig that pretty quickly (1
> hour) based on what Blake wrote for the /plugin-info thing if anyone
> is interested.  Essentially it'd involve creating a url_handler
> callback chain which gets called at the beginning with the url and
> environment data.  Then the first registered function that can handle
> the url in the chain handles it and the chain stops.

Then you need to change some of the code in startup() and run(), maybe
combine some of the URL discovery, and each url_handler either returns
an entryList or do something and stop there (You may want head.html and
foot.html templates to run). So this is not a hack, more thoughts have to
be thunked before doing something big like this.

> Blake mentioned in a previous email a bunch of issues with this namely
> what happens when two things want to handle the same url.  I think to
> that I would say we're designing a pretty small system here and while
> that's something to think about for functional completeness, I don't
> think it's an issue we really care about (I know I don't care :) ).
> If a plugin writer has this problem, then they can always register
> themselves with the callback chain with a higher priority and thus
> pre-empt other handling function.

I can possibly classify some types of URL that url_handler would want to
process.

1. Category, directories of your blosxom blogs
2. Permanent entry links
3. Year, month, date. These are standard blosxom stuff
4. ?q= for queries
5. ?flav= for flavours
6. ?q=xxx&flav=xxx both
7. /plugin-info or whatever plugins that 'hijacks' item 1 and 2
8. Some form of comments or feedbacks.
9. All of the above are GET data, what about POST? PUT or DELETE?
10. 1 2 3 7 8 should usually work together with 5

Of course a default handler could do what pyblosxom is handling now,
which is 1 2 3 5. Not that easy I think, or there could be several
handlers for such things that worked in order.

 * Handler for entries (Entries and or feedback outputs)
 * Handler for events (feedback inputs, search queries)
 * Handler for special plugins like /plugin_info
 * Handler for actions (redirect after feedback inputs?)

> I think when a function handles the url, it's building the whole page
> and that's the end of our run.  I can't think of a use case where we'd
> want two different functions to handle the url and do different
> things.

Thought of already, above :)

> I vaguely remember there being other issues, but I can't remember what
> they were.

Not issues, just proper implementation needed.

-- 
Regards: Wari Wahab
Senior R&D Engineer
Celestix Networks
http://www.celestix.com/