[Bese-devel] Re: more feedback

[Marco Baringer]

On Venerdì, apr 2, 2004, at 20:34 Europe/Rome, Jan Rychter wrote:

> I've updated ucw and yaclml to the latest versions.
>
> Hmm. interesting. I have figured out at least one cause of my problems:
> I have configured my Apache as:
>
>   <Location />
>   SetHandler lisp-handler
>   </Location>

ucw assumes (implicitly) that the only requests it will get are 
requests it should handle, it doesn't know how to serve "regular" files 
and i don't really think i want to teach it to. though i may need to at 
least teach aserve to distinguish.

you'll notice that the recent entry points are all mapped to urls 
ending in ".ucw", this not to remind everyone that they're using ucw, 
this is so that you can do

<LocationMatch "\.ucw$">
   SetHandler lisp-handler
</LocationMatch>

And let apache deal with serving icons, static files and style sheets.

> BTW, a quick manual of how to deal with crashed/broken UCW sessions
> would be extremely helpful. How do you do it? I feel major pain having
> to manually kill my *inferior-lisp*, *slime-repl* and quite often also
> *sldb* buffers along with my lisp...

i generally don't kill those. although adding a slime-quit function to 
slime wouldn't be a bad idea.

i usually start ucw from the terminal and then immediatly attach a 
slime to it. (i've even got a script which starts ucw, starts a slime 
server and then does a emacsclient -n -e '(slime-connect ...)', maybe 
i'll put this script into the main repo). When things go wrong i use 
M-x slime-list-threads to attach a debugger to the proper thread, i try 
and figure out what went wrong and either abort the request via the 
server-error restart, or re-call the action via the retry restart.

of course, this is what i _would_ do, at the moment it seems that my 
slime-attach-debugger code is broken. :(

--
Marco
Ring the bells that still can ring.
Forget the perfect offering.
There is a crack in everything.
That's how the light gets in.
	-Leonard Cohen