httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject Re: [PATCH] ap_add_filter
Date Thu, 17 Aug 2000 23:52:45 GMT
On Thu, 17 Aug 2000, Greg Stein wrote:

> On Thu, Aug 17, 2000 at 09:22:03AM -0700, wrote:
> This is goodness. Can we add this and defer the FIFO/LIFO thing? I still
> think that the LIFO is not going to solve the basic problem: filter
> insertion time can be non-deterministic, thus we can never insert them in
> the proper time-based order, therefore we need some internal classification
> of the filters to ensure proper ordering (among classes).
> I've also seen that the connection/transport filters probably operate more
> like a prepend-list, while the content filters are more like today's
> append-list.
> I think that the switch from append-style to prepend-style is based on the
> experience of just the chunking filter. What happens with the next filter?
> And after that? Let's keep the mechanism as we originally planned until we
> see more examples and have more details on the model.
> There is also the insert_filter hook ordering vs filter ordering. Reversing
> these irks me a bit. This flipped model, in conjunction with the
> non-deterministic nature of inserting filters makes me say that we shouldn't
> flip the sequence (now).

The insert_filter thing is wrong.  Most modules won't be using it, because
it can only deal with those things that are determined in the config
file.  That is what I am trying to get across here.  We are wrong

> Summary: can we add the "next" stuff and defer the prepend/append thing
> until we have more experience with actual filters?
> IOW: -1 on the prepend/append. +1 on the "next" param to add_filter.

That doesn't solve the current problem.  Here's the way I see it.  We have
something that doesn't work, that we designed before ever having any
experience with actually writing filters.  And, we have another solution
that does work, and is based on the experience of writing a filter.  One
of the people supporting the new method (Roy) has more experience with
filters in general, AFAIK.

What is the argument for not switching now, and avoiding releasing an
alpha with a serious hack that won't work for any module other than


Ryan Bloom               
406 29th St.
San Francisco, CA 94131

View raw message