Re: Independent numbers of columns for matching groups?

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

Adam Spiers wrote:

> This may have been discussed before and I missed it, but would it be
> possible to changing the group_matches code so that the number of
> columns is calculated individually for each matching group?
> 
> An example of why I ask is _rpm; sometimes it offers the following
> completion groups:
> 
>   - RPM packages
>   - RPM package files
>   - ftp prefixes
> 
> RPM packages are always long, which forces the whole output into a
> long single-column list, even though the RPM package files section
> could easily fit in multiple columns and save screen space.

Yes, I saw this, too, sometimes. The only problem I have with changing 
it right away is that for groups that don't differ so much I prefer
having all columns aligned (well, another problem I have with this is
that it would make {i,comp}listmatches() more complicated, but that
shouldn't be a reason not to allow this). So maybe we should make this 
an option are think about some clever way to make it automatically
decide if the columns should be aligned for all groups or not. That
would be: if the minimum and maximum number of columns for all groups
differ only by one (or two? or by `max / n'?), align all groups.

Maybe we'll just have to play with this (and hence implement it).

This gives me the oppotunity to repeat my question about the other two 
possible enhancements: Andrej's suggestion to allow matches to be
layed out in a row-first order and my old LIST_PACKED (see 7590).
Interesting enough?

Bye
 Sven


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