httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From r..@covalent.net
Subject Re: Add hook to get server read.
Date Thu, 11 May 2000 19:56:57 GMT

> Cool. I'm out of town from Friday late morning until Sunday afternoon. If
> you get it done before I leave: sweet; if not, then I'll review when I get
> back (whether you've held it or checked it in... I can find it :-)  I'd
> prefer if you held it, but it's your call, of course.

Should be done RSN, but of course that assumes no interruptions from other
stuff.  :-)

> 
> >...
> > > > +static const char * ap_build_cont_config(ap_pool_t *p, ap_pool_t *temp_pool,
> > > 
> > > What does this function do? Can we get a one-line comment or something? It
> > > seems to reproduce a good amount of code from elsewhere. It looks like it
> > > sucks up lines from a config file, creating directives. But it doesn't
> > > process the EXEC_ON_READ directives like the "real" processing stuff.
> > 
> > It builds a sub tree for the directives within a container.
> 
> Understood, but as I mentioned... it seems like we already have code for
> that purpose. And there is the lack of EXEC_ON_READ; if that is
> intentional, then maybe we can refactor the "build a container" code to
> take a flag on whether to obey EXEC_ON_READ.

Nope.  The other code always reads until the end of a file.  We want to
stop the read at the end of a container.  This seems like a minor thing,
but it really isn't.  I'll look into combining them, but I don't see much
hope there.  As far as not checkinf for EXEC_ON_READ, that is an
oversight.

> 
> I can see at least two types of execution:
> 
> 1) build tree and call EXEC_ON_READ
> 2) throw out a container (due to a filed conditional like IfModule)
> 
> I'm confused about the build with no EXEC_ON_READ.

Yep, you are correct, this will be fixed in the next patch.

> > Yes, execute now returns a sub_tree which can be put back into the
> > tree.  This is how the code to get rid of the branches for ifmodule and
> > ifdefine works.
> 
> I think you mean "when ifmodule/ifdefine want to *include* the branch,
> then they return the sub_tree." Correct?

Since I'm having a hard time parsing both your sentance and mine, I'll
start over.  If IfModule/IfDefine evaluate to true, a sub_tree is built
and included in the server config.  If they evaluate to false, we soak up
that container and move on.

> I understand that, but I don't see the other side. The part where we set
> up the integer and then examine it.

Ignore it, it's going away now.

> 
> > > Can there be a comment somwhere that simply states "an EXEC_ON_READ
> > > returns a boolean on whether the interior block should be processed?" or
> > > something like that. It seems like that is what is happening.
> > 
> > This is where things can and should be changed.  This will be obvious in a
> > while.  :-)
> 
> Okay :-)
> 
> Coolness. I look forward to the new set, to see your "new idea". I presume
> it cleans/clarifies :-)
> 
> But... no rush! I'm just Mr Peanut Gallery right now :-)

Yep.

Ryan

_______________________________________________________________________________
Ryan Bloom                        	rbb@apache.org
406 29th St.
San Francisco, CA 94131
-------------------------------------------------------------------------------


Mime
View raw message