[Gnu-arch-users] Re: Preliminary Arch Cache available

[Stefan Monnier]

> I gather, for example, that this code essentially caches file-fetches
> from archives.   Well, that could be done as well by using proxies to
> read those archives, with absolutely no changes required for arch at
> all.

I don't know of any caching proxy for sftp.
Most http proxies will not cache https requests.
I suspect that the same might be true for webdav (to I must admit to
ignorance here).

All those caching proxies have no idea that the data is actually immutable,
so they'll very often flush the data frmo the cache unnecessarily, or
they'll needlessly contact the server to check freshness.  Inversely, they
sometimes will serve you stale data (so they'll say "nope, nothing new here"
even tho you know that someone just committed a new revision).

It's all in all basically incomparable: On the one hand, you have a hackish
heuristic that works 90% and on the other you simple and efficient system
that just does what you want.


        Stefan