[ccache] Implementing a Read-only HTTP CCACHE_DIR

[Richard Maw]

Hello, I've not written to a mailing list before so tell
me if I have done anything wrong.

For a project I am working on we will be rebuilding a lot of programs,
so we are using ccache for that, however the idea is that a lot of
people may be building the same thing, so we want to make the ccache
objects publically available, especially since we could be building on
machines that have better IO speeds than CPU speeds.

To this end we are considering adding the ability to look up the objects
on an http server.

I have spent half a day looking at the source code and planning what to
do, but you know more about ccache than I do, so I was wondering if
anyone else had tried this kind of thing before, or if there is some
fundamental problem with the approach that I haven't considered.

The the plan is that after the hash has been calculated it checks
whether the object is cached locally, if it hasn't it tries the http
server specified by an environment variable.
If the server has the object then ccache downloads the object and uses
that. If it hasn't it carries on and compiles it.