Zsh Mailing List Archive
Messages sorted by:
Re: About menu selection (again)
- X-seq: zsh-workers 17557
- From: "Bart Schaefer" <schaefer@xxxxxxxxxxxxxxxx>
- To: Sven Wischnowsky <wischnow@xxxxxxxxx>, zsh-workers@xxxxxxxxxx
- Subject: Re: About menu selection (again)
- Date: Sat, 17 Aug 2002 17:26:39 +0000
- In-reply-to: <15708.60068.71382.607117@xxxxxxxxxxxxxxxxxx>
- Mailing-list: contact zsh-workers-help@xxxxxxxxxx; run by ezmlm
- References: <3D56CD6E.mailMX1LBVZ3@xxxxxxxxxxxx> <1020812060148.ZM28748@xxxxxxxxxxxxxxxxxxxxxxx> <3D5906BF.mail17S21TJ8M@xxxxxxxxxxxx> <1020813150050.ZM30758@xxxxxxxxxxxxxxxxxxxxxxx> <3D5954F7.mail1L21A9GN@xxxxxxxxxxxx> <3D595620.mail3I11T8QF@xxxxxxxxxxxx> <1020814022702.ZM31935@xxxxxxxxxxxxxxxxxxxxxxx> <3D5CEB3E.mailV211HY5E@xxxxxxxxxxxx> <15708.60068.71382.607117@xxxxxxxxxxxxxxxxxx>
On Aug 16, 2:05pm, Sven Wischnowsky wrote:
} DervishD wrote:
} > Since complist needs compctl to work, could it be added as a dependency?
} That would be wrong because complist doesn't need compctl. It only
} `needs' some form of completion to be useful and the new completion
} system is only built on top of the more basic `complete' module.
The trouble is that the zsh/complete module is not a form of completion;
it's a collection of tools for building forms of completion. It's useless
unless `zle -C' has been run and that widget calls `compadd'.
So the test in zle_tricky.c:docompletion for module_loaded("zsh/complete")
is woefully inadequate. We need to at least test for `zle -C'.
Could we do something along the lines of shuffling the hookdefs so that
the completion hooks are not defined until `zle -C' runs the first time,
even if the zsh/complete module has been loaded? And then test for the
presence of the hook rather than the presence of the module, and load
compctl if the hook doesn't exist?
Bart Schaefer Brass Lantern Enterprises
Zsh: http://www.zsh.org | PHPerl Project: http://phperl.sourceforge.net
Messages sorted by: