Re: using modules in completion functions

[Sven Wischnowsky <wischnow@informatik.hu-berlin.de>]

Oliver Kiddle wrote:

> While looking through the manual, I noticed the disable-stat style is
> used for cvs completion when deciding whether to load the stat module.
> For completing file descriptors, I had taken a different approach where
> the stat module was used if already loaded but if not, ls is used. I
> would expect that it will be a fairly common situation that we will
> have where a completion function can make good use of a module but can
> cope without it and we have to decide whether or not to use it. It
> would therefore be a good idea if we agreed on a common way to handle
> this.
> 
> What I would suggest is a style named something like use-module which
> can be set to one of three values which would correspond to:
> 1. always use the module, loading it if necessary
> 2. never use the module
> 3. use the module but only if it is already loaded
> where the third should probably be the default
> 
> The name of the module could be placed somewhere in the context. Where
> would it make sense to put it - the argument, as a sub-command or
> elsewhere?

I'd think: as the tag, if at all.

> If it can't go in the context, I suppose we could have
> separate styles such as use-module-stat but that wouldn't be so good in
> my opinion.

Or use only one style and the value gives the names of the modules to
use with some kind of token to distinguish between `load always' and
`use if loaded'. If you really want that, because, if the module is
already loaded, why not use it unconditionally? I mean, it should be
faster than any alternative, otherwise using the module doesn't make
sense anyway. In that case the style could be named `load-module', if
course.


Bye
 Sven


--
Sven Wischnowsky                         wischnow@informatik.hu-berlin.de