[Tilecache] trac ?

[Attila Csipa]

On Tuesday 22 January 2008 04:03:28 you wrote:
> The implementation seems sane in general, though I'm not sure I like the
> extra cost of doing an md5sum should be turned on by default, but rather

There are of course 'cheaper' ways to produce the identifier - I used md5 bc 
of future plans of caching tiles that don't have a storage that can provide a 
date. I can check the date if it's a file, but if it's in memcache I'd have 
to check for content to make sure it's really the same (this is not as 
expensive there as it seems, the hash can be cached, too, awkward as that may 
sound). This patch does not provide any extra functionality for memcached 
based datastores at the moment so I'll swap it with something simpler for 
now.

> an option. Also, before the patch goes into trunk, the other handler
> functions will need to be modified to support the change as well.

OK, just didn't want to start the roof until I have some feedback if my 
fundaments are actually sound.

> So, if the patch were to:
>  * Add support for the other handler functions
>  * Make the caching header stuff a service-level metadata option
>  * Ensure that the default behavior of dispatchRequest continues to be
>    to return only two args, and only return 3 if the service-level
>    metadata option described above is 'turned on'
> Then I would probably take it without much question. As is, the general
> principle is sound, but I'm not convinced that the implementation is
> complete enough to merit pushing it into trunk yet.

I'll resubmit the patch with this in mind when it's ready.