On Friday 11 March 2005 12:37, Oliver Kiddle wrote:
>
> > Completion for SVR4 is not quite right - it should complete
> > signal names only after -k, but it seems to be too much work.
>
> It isn't too hard and it is always very annoying to break file
> completion (as this does for Solaris). Does this patch do the right
> thing?
>

How does it differ from current completion?

I actually meant - signal names should be completed only if -k is present 
(because they are meaningless without). But currently it is possible to 
implement only using ->state and it seemed too much work for such small 
function.

Will "prerequisite options" be generally useful? I.e. something like

[-a]-b:....

with semantic "-b is offered only if -a is already present". I suspect this 
may have simplified a number of functions.

-andrey