Re: The problem with mod_dav..
Nick Kew wrote: However, if you're suggesting improvements to mod_dav, you should really be on [EMAIL PROTECTED], which is where any changes would have to be decided And I was thinking about posting there as well.. Meh. I'll start hammering out a version that does what I want and see what [EMAIL PROTECTED] says :)
The problem with mod_dav..
.. is that if you want to add further methods then mod_dav itself needs to be modified. I spoke briefly with Nick on IRC a few days ago and the inference appeared to be that I was right and things needed to change with mod_dav. Or I needed to basically emulate half of the behaviour found in mod_dav already. Which seems to be pointless. Right now I've only got one /partially/-sane idea whereby mod_dav gets completely ripped to pieces and the provider api gets replaced by a new one where a provider module can itself register functions against methods i.e. something like dav_hook_reg(METHODNAME, functionpointer, ...) which would register an entry in a linked list, for that provider, within mod_dav. The mod_dav handler would then traverse this list and execute the relevant hook(s). Can anyone see any big flaws with that? Other than breaking every dav provider? Regards, Karl
Re: Introducing mod_enigform.
Arturo 'Buanzo' Busleiman wrote: I'd appreciatte input on what kind of configuration it would be nice to have. So far I thought of Order/Allow/Deny, but I'd like it to be more flexible. If there's anyone who'd like to get hands on the code, let me know. I also don't know if this code should be hsoted on apache.org's CVS servers, or what, as I plan to release this to the Apache foundation, so Apache becomes the first HTTP server to support the upcoming IETF Draft that all this is about. This might be off at a bit of a tangent, but I'd love to test this out and I'd be interested in seeing some sort of provision for redirection or something, if the signing isn't available. Possibly a little out of scope as this is achievable through mod_rewrite. What I'd really love to see is support for mod_dbd, etc. so that keys could be stored in a database and yanked from there. Also off at a giant tangent, is there any plans for a signed response in the specs (I assume this would require a fully buffered response, which would be rather expensive)? If so, any plans on this module supporting that? Regards, Karl