Re: Post Processing

On Thu, Dec 18, 2003 at 09:48:46AM +0100, Dongen, Jeroen van wrote:
> Jacob et al.,
> 
> Option 2 seems the most practical indeed - although option 1) would get a
> vote for consistency I guess. Option 3 is too much work for too little gain.

And option 2 is now checked in.  The response is stored for the
duration of that hook in sessioDict['RESPONSE'].

The error handling (at the moment, there is none) is probably not
right.  If an error occurs, it should be caught and we should return a
status 500.  At the moment, however, if you have division by zero in a
filter, or your HTMLParser barfs on a page and you don't catch it, the
error will simply get raised and the client won't get a 500.  The
problem is that the web service doesn't offer "500 protection" for
anything, as folks may discover if they use a crazy rewrite rule or
are adding hooks of their own; the 500 protection in skunkweb is
implemented in templating.  That may be the wrong place for its
primary residence.  We may need a HandleServerError hook in web.

> BTW: has the doc-sprint happened and if so, what's the verdict?

It was a more of a sprintlet, as I only had a couple of hours, which
meant I rewrote a few paragraphs.  We decided to concentrate on the
api section for now, and I did some work on ae.txt.  Holidays are upon
us, but I'd like to have weekly doc sprints, even if they are short,
to keep the fuse lit.  

j

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.3 (GNU/Linux)

iD8DBQE/4eE4uqamFyFXXLIRAuwvAJ42oRtaTwRfKLFlxsk/y/ihhbDP/gCffGve
PHmB/VNtOCJnt1jqtAcIpf0=
=mKSz
-----END PGP SIGNATURE-----