Zsh Mailing List Archive
Messages sorted by:
Reverse Date,
Date,
Thread,
Author
Re: Caching between matchers
- X-seq: zsh-workers 16312
- From: Bart Schaefer <schaefer@xxxxxxxxxxxxxxxx>
- To: Borsenkow Andrej <Andrej.Borsenkow@xxxxxxxxxxxxxx>, Zsh hackers list <zsh-workers@xxxxxxxxxx>
- Subject: Re: Caching between matchers
- Date: Sat, 8 Dec 2001 17:51:31 +0000
- In-reply-to: <1007824073.29799.1.camel@xxxxxxxxxxxxxxxxxxxxx>
- Mailing-list: contact zsh-workers-help@xxxxxxxxxx; run by ezmlm
- References: <1007824073.29799.1.camel@xxxxxxxxxxxxxxxxxxxxx>
On Dec 8, 6:07pm, Borsenkow Andrej wrote:
} Subject: Caching between matchers
}
} When completion generates list of matche on-the-fly (like installed RPMs
} or list of available RPMs in urpmi completion) it currently does it for
} every matcher that is tried. If list is long (urpmi has ~3000 packages
} in database) or needs some time to build (esp. when forking off) it may
} take quite a bit of time.
This is why the caching mechanism (_store_cache, _retrieve_cache, and
_cache_invalid) was introduced; _urpmi should be doing its own caching.
} Because the list obviously is not changed between matchers, is it
} possible to reuse it?
This isn't precisely true. Many completers test $_matcher_num, which is
set by main complete, and do nothing if it is greater than 1. On the
other hand, _path_files tests whether the *last* matcher is being tried,
and may add more path expansions to the list in that case.
} Actually, is it possible to traverse matchers list internally in
} compadd?
I don't think so, because the matchers may be traversed multiple times
(e.g. the _prefix and _ignored completers re-traverse the matcher-list).
--
Bart Schaefer Brass Lantern Enterprises
http://www.well.com/user/barts http://www.brasslantern.com
Zsh: http://www.zsh.org | PHPerl Project: http://phperl.sourceforge.net
Messages sorted by:
Reverse Date,
Date,
Thread,
Author