Re: ApacheModulePerl.dll

[Doug MacEachern <do...@opengroup.org>]

Dean Gaudet <dg...@arctic.org> wrote:

> 
> 
> On Mon, 28 Jul 1997, Doug MacEachern wrote:
> 
> > max_requests_per_child
> 
> (I'll comment on the rest later.)
> 
> I'm thinking we should have a struct global_conf where all these global
> config values are stored, and then export that.

Good idea.
 
> Oh yeah, I'm curious why you use basic_http_header instead of
> send_http_header.  Maybe our abstraction is wrong there. 

It's only used if perl version is <= 5.004 and the CGI::Switch module
is used.  With 5.003, it's not possible to redirect STDOUT and STDIN
to the client, so this subclass did pulls a few tricks underneath in
CGI::Apache.  One trick it pulls is to call basic_http_header() in
the header() method, send_http_header() would break scripts here
because it sends the terminating \r\n  It's no longer needed, now that
Perl has tie'd filehandles, but these CGI:: modules ship along with
CGI.pm in the Perl distribution, some folks are still using it w/
5.003 and 5.004.  At some point, CGI::Apache's use of
basic_http_header will go away all together.  Actually, this would be
a good time to start pushing for this, so don't API_EXPORT
basic_http_header().

-Doug

Re: ApacheModulePerl.dll

[Dean Gaudet <dg...@arctic.org>]

On Mon, 28 Jul 1997, Doug MacEachern wrote:

> Dean Gaudet <dg...@arctic.org> wrote:
> 
> > On Mon, 28 Jul 1997, Doug MacEachern wrote:
> > 
> > > max_requests_per_child
> > 
> > (I'll comment on the rest later.)
> > 
> > I'm thinking we should have a struct global_conf where all these global
> > config values are stored, and then export that.
> 
> Good idea.

A problem I just realised was that this makes it hard to make changes to
the globals without breaking binary compatibility.  But... that's a pretty
normal occurance for us.

> > Oh yeah, I'm curious why you use basic_http_header instead of
> > send_http_header.  Maybe our abstraction is wrong there. 
> 
> basic_http_header will go away all together.  Actually, this would be
> a good time to start pushing for this, so don't API_EXPORT
> basic_http_header().

sounds like a good plan :) 

Dean