Re: [Gnu-arch-users] Re: Arch Cache & cached archives

[Tom Lord]

    > From: Stefan Monnier <address@hidden>

    > > I think that caching vs. memoizing should (at least ultimately) be a
    > > finely tunable parameter.

    > Could someone explain to me what is the difference between caching and
    > memoizing here?

Memoizing never forgets.  Caching does.

Another way to say it: a memo is like an infinitely large cache (no
entries are ever displaced).

A revlib is a memo in the sense that nothing in arch automagically
removes revlib entries.   It should be more like a cache but some
design work (this thread included) needs to happen first.

Some memos, like revlibs, allow entries to be removed by hand --- but
it's still useful to think of them as memos (since we don't like
adding "you need to do this part by hand" steps).


    > > One could implement cached archives without changing arch in the
    > > slightest.

    > I don't know to what trick you're referring to.

A proxy server for archive access.

    > The closest thing I can think of is an http caching proxy, but
    > that will only work for some archives.

Do you think it would be hard to write a generic proxy that could
handle arch's limited use of FTP, SFTP, and WebDAV?   Sure --
independent of arch that would be a very hard problem but given arch's
referential transparency, it should be trivial.

-t