httpd-apreq-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Max Kellermann <>
Subject Re: PATCH [multi-env]: inline functions, apreq_initialize()
Date Mon, 07 Feb 2005 20:22:59 GMT
On 2005/02/07 20:46, Joe Schaefer <> wrote:
> I think we should just remove apreq_param_brigade.  The
> general issue with exposing a bucket brigade is that
> there are side effects which happen during file-bucket reads.
> Users that want to read  data from the brigade should
> make a copy of the brigade first, and then successively
> read+delete the buckets.

I don't understand that yet - how do you want users to access the
upload contents? Only by installing an apreq_hook_t?

Only removing apreq_param_brigade hides nothing, as there is still
apreq_param_t.upload. If that is also removed, the user cannot
distinguish between uploads and normal parameters.

> I don't mind making "check" and "test" synonyms (perl guys prefer
> "test").

automake prefers the target "check". The perl scripts may still be
named "test"...

I'd suggest to make the target "all" build everything, including the
tests. The user can enable or disable the test suite build with
"--enable-test". What do you think?

> The ambitious plan for the t/ would be to remove CuTest.*
> and replace it with an Apache::Test-based C framework, modeled 
> around independent C executables that emit TAP (the "ok/fail" 
> protocol).  We should add an svn:external to the /asf/perl/Apache-Test 
> repos and include that in our releases, just like modperl does.
> That way we can remove the Apache::Test dependency from our 
> prereqs.


> > - move each parser into its own .c file?
> A good idea; but there's some private interaction between the 
> header parser and the mfd parser that I think we'd need to
> look closer at.

"some private interaction" should be defined and converted to a
documented interface ;) I'll work that out.

> > - rename apreq_env_handle_t to apreq_t?
> I think it should be apreq_handle_t, to complement apreq_module_t.

ok for me.


View raw message