httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ryan Bloom <...@covalent.net>
Subject Re: mod_include vs file descriptors
Date Tue, 13 Nov 2001 17:39:08 GMT
On Tuesday 13 November 2001 09:39 am, Cliff Woolley wrote:
> On Tue, 13 Nov 2001, Ryan Bloom wrote:
> > I should clarify that just destroying the subrequest should be
> > perfectly safe. Have we tried that yet.  In the past, we would have
> > had a problem with data surviving the death of the request, but all
> > buckets should have cleanups now, so that shouldn't be an issue.
>
> Hang on... you do have to explicitly setaside all the buckets you want to
> keep before destroying the subrequest and tell it which pool you want them
> setaside into.  Buckets don't have cleanups (well, pool buckets do, but
> none of the others do), they have setaside functions.

We will have done that already.  This may get complex, but follow my logic,
and point out anything I missed.

We are talking about destroying the sub request.  By the time this is done,
we have already passed all of the data from the sub-requests filter stack to
the original requests.  Either all of the data was written to the network, or
it was set-aside.  If it was written to the network, we can ignore it.  If it was
set aside, then it was done so in the original request, so we passed r->pool
or c->pool to the setaside function.

I wasn't clear in my original message, but the result should be the same.

Ryan

______________________________________________________________
Ryan Bloom				rbb@apache.org
Covalent Technologies			rbb@covalent.net
--------------------------------------------------------------

Mime
View raw message