httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "William A. Rowe, Jr." <wr...@lnd.com>
Subject RE: Config tree code finished.
Date Wed, 26 Apr 2000 14:34:12 GMT


> -----Original Message-----
> From: Greg Stein [mailto:gstein@lyra.org]
> Sent: Wednesday, April 26, 2000 9:34 AM
> To: new-httpd@apache.org
> Subject: RE: Config tree code finished.
> 
> 
> On Wed, 26 Apr 2000, William A. Rowe, Jr. wrote:
> > > As I've outlined, this is not possible. The directive is 
> not available
> > > during the read process. Therefore, you cannot flag the 
> > > <perl> directive to suck in a text block.
> > 
> > Why not?  The only XMLish notiation in the current structure (that
> > I have found) are tag /endtag blocks.
> 
> The module has not been LOADED during the construction pass. 
> Therefore,
> you cannot have any information about parsing its directives.
> 
> Sure, you can deal with directives in general, but you cannot say "the
> <perl> directive has a flag that indicates the contents 
> should be treated
> as an unparsed block of text" because the code that defines 
> <perl> hasn't
> been loaded.
> 
> (unless you accept my hook design; specifically #2 -- hook in at node
>  construction time)

I did look at that and agree.  We need to treat -unrecognized- directives
as unparsed argument lines, and -unrecognized- block directives as raw
text blocks, and then run an entire second pass on the tree with all the
macro expansion, dso module reparsing and other non-static behavior. 

Mime
View raw message