Alvaro Lopez Ortega wrote:
> Directory / {
> Handler cgi {
> scriptalias /home/esj/akasha/x.sh
ErrorHandler on
> }
> }
You forgot to tell the server you want this handler to manage the
error responses as well.. :)
booo hisss that should be the default... :-)
in either case, the documentation should contain something like...
by default, Cherokee will use a server provided warning if there is a
400 or 500 class error message. For example, a handler detected 404
error will return the standard "page not found" message. It is possible
however to customize that by either using the 404 redirection handler.
(URL here). But in the case of the CGI program such as the one shown in
the configuration below, the CGI program could return its own error page
just like any other web page. In order to do that one must turn on on
the error handler feature as shown below. If the CGI program does not
return anything (what will happen?)
Directory /example {
Handler cgi {
Scriptalias /usr/lib/cgi-bin/example.cgi
ErrorHandler on
}
}
> n.p. is CHEROKEE_TRACE a compile option?
It is gonna be one of the new features in 0.5.0. You have to
execute ./configure with the --enable-trace flag, and it will
compile the library with extra debugging information.
okay, let me know when you want me to drag something out of svn.
If you want to trace something, you only need to set the
CHEROKEE_TRACE variable.. for example:
- CHEROKEE_TRACE="handler" /usr/sbin/cherokee. This will print
debugging information about what are the handlers doing.
.. the only weak point is how to figure out what the possible values
of the environment variable are, but I will try to work it out.
I went through the same thing you are with camram and its log based
debugging. I haven't come up with a good solution. Part of the problem
is that you have multiple ways of slicing the information stream. For
example, you may want to see everything coming from one particular
function/method and nothing else. Other times you want to see all of
the priority one messages (i.e. warnings, exceptions, etc.) so I haven't
made it work right yet. :-) but I do have hopes.
create a list of all the symbols, each log entry uses a key into that
list. If the specified list entry matches the environment variable,
that's your trigger. This does make it a little more difficult in that
you need to maintain an enumeration in parallel with the list and
hopefully they won't get out of sync. That seems like it might be
manageable. On the other hand, it's been about 10 years since I've done
any significant C/C++ coding
--- eric
_______________________________________________
Cherokee mailing list
Cherokee@lists.alobbs.com
http://www.alobbs.com/cgi-bin/mailman/listinfo/cherokee
