Re: pyblosxom 3000

On Fri, 14 Jan 2005 10:30:40 +0100, Steven Armstrong <[email protected]> wrote:
<snip>
> 
> As I'm not working fulltime atm and therefore have enough free time I
> guess I could do this. Can someone describe what exactly the contrib
> maintainer is supposed to do?
> 

Gotta talk to Will on that one.

> >
> > 5) Comment Administrator - I have one that works, more or less, except
> > it can't change the dates on comments. This is because I want to
> > change the comments module to use the date field inside the comments
> > file instead of the time embedded in the filename.
> 
> Cool, I was thinking about hacking something like that together.
> Courious how you did it.
> 

I did it quick & dirty; I'm working on it today, hopefully it'll get nicer.

> <snip>
> >
> > 7) mod_python / WSGI integration - Didn't somebody submit a patch/some
> > patches that do this? Are we going to try and integrate it for 1.2? I
> > think it's a good idea.
> >
> > A quick search shows
> > http://sourceforge.net/mailarchive/message.php?msg_id=10250388 .
> > Steven, you still hanging around here? Want to step up for the effort
> > to patch it in?
> >
> > I'd like to start running this on my personal blog.
> 
> Yeah, I'm still here ;-)
> A few issues/thoughts I bumped into while implementing the
> WSGI/mod_python stuff:
> 
> There are quit a few modules/plugins that directly access sys.stdin and
> sys.stdout. This turned out to be a problem for the WSGI/mod_python
> approach.
> For example, mod_python requires the content-type to be set in a special
> way on the mod_python request instance. If a module/plugin writes it
> directly to sys.stdout it messes up everything. Similar if a
> module/plugin reads directly from sys.stdin.
> 

This is the type of change I think we need to get started on. Do we
want to break compatibility for some modules for the next release? I
say yes, and that we should start breaking what we need to soon, so
that we can get a solid 1.2 api in place.

> Another problem I had was with how the outputstream/renderer stuff
> works. For example in my nospam plugin I implemented the cb_handle
> callback to generate and write out the human-verification-image.
> As the renderer is not set at this stage, the only way to add headers
> and write data was to directly use stdout. This again fucked up mod_python.
> 
> I think it would be a good idea to split the outputstream handling from
> the renderer. Let the renderer generate the content and use a seperat
> object to handle all output related stuff (e.g. add headers and write
> the response).
> 
> Something like:
> response = request.getResponse()
> response.addHeader('Content-Type', 'text/plain')
> response.write("Hello World")
> 
> Can we change PyBlosxom to work like this? Would make my life easier.
> 

I like it; want to work on it and submit a patch? I don't think it'd
be that hard to redirect all output through a response object.

Peace
Bill Mill
bill.mill at gmail.com


-------------------------------------------------------
The SF.Net email is sponsored by: Beat the post-holiday blues
Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt