[Tilecache] Caching identical tiles WAS Re: http 304 Not Modifies responses ?

[Attila Csipa]

On Wednesday 02 January 2008 16:15:35 Christopher Schmidt wrote:
> > As a future possibility I was thinking of using the etag as a base for
> > transparently supporting/caching/replacing common (pure black/white)
> > tiles without actually loading them (there was talk about this earlier,
> > although I'm not sure if here or on the mapserver-dev list).
>
> I think that we should talk about this is you're interested before you
> implement it: I have some vague implementation notions that I'd like to
> bounce about before we do too much coding on it

I have several ideas on this, but all just in idea phase, I wasn't thinking of 
implementing them right away. The main idea is to cache repetitive tiles 
(should not matter if they are white or blue, etc). As a sort of brainstorm, 
what I had in mind was:

- use 307 Temporary redirect to point the client to known repeat tiles (which 
can then be 304-ed)
- do not write such files to the filesystem - if a tile turns out to be 
monocolor, mark it somehow (special filename ? softlink ? added to a list ?)
- if we know it's a monocolor tile from the above point, it should be faster 
to generate it in memory and just pass it through
- perhaps replace large monocolor tiles with 1x1 counterparts

some of these can yield unexpected results and breakage, but still, more on 
the line of just thinking out loud...