On Monday 21 October 2002 09:17 am, Robin Berjon wrote:
> Tod Harter wrote:
> > On Monday 21 October 2002 05:06 am, Robin Berjon wrote:
> >>Todd W wrote:
> >>>( I use Apache::StatINC ).
> >>
> >>It looks as if My::WeatherTaglib is not inheriting from
> >>Apache::AxKit::Language::XSP like it should. If you add that does it
> >> work?
> >
> > The other possibility which I've seen is that in certain circumstances
> > (and I am not sure why, I never really investigated too much) a Taglib
> > can fail to compile for any old ordinary reason like some other module it
> > uses is missing, and AxKit/mod_perl will happily go right along and do
> > its thing with no visible sign of error until you try to serve the page
> > in question. Then you get something like this message about not being
> > able to register, just as if the taglib had never loaded. Where did the
> > error message go? Who knows...
>
> That's a common artefact of using Apache::Reload or Apache::StatINC, and
> the behaviour you describe could indeed be what's happening here, I've
> seen it myself. The reason for this is (I think) that those modules use
> evals (thus the server doesn't crash upon error) and that once they've
> failed to load a module once, they'll leave it out and won't ever
> reload it.

I expect your right Robin. I pretty much gave up on both Reload and StatINC 
long ago. They're kinda handy for simple stuff, but they never really seemed 
very reliable when doing XSP anyhow... Besides, I tend to find that AxKit's 
stylesheet cache gets confused pretty easy anyhow, so restarting apache is a 
highly ingrained habit...

(actually I think the precise problem is if you XINCLUDE data during SAX 
processing then very strange things happen). One thing I wonder about, how 
come XSP doesn't support XINCLUDE? It just seems to ignore it.
>
> While on the topic, I'd recommend Apache::Reload over Apache::StatINC,
> especially as you'll need it if you want your "Certified 100% Pure
> MSERGEANT Solution Provider" label.

LOL. 


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to