Re: PATCH: _diff (new), _prcs (upgrade)

[Alexandre Duret-Lutz <duret_g@xxxxxxxx>]

>>> "Sven" == Sven Wischnowsky <wischnow@xxxxxxxxxxxxxxxxxxxxxxx> writes:

 Sven> Alexandre Duret-Lutz wrote:

[...]

 >> My opinion is: we should write $(XXX) if we are in the completion
 >> function for XXX, and $(command XXX) otherwise.

[...]

 Sven> But then a hackishly written function may still break everything (if
 Sven> you have a function for diff that actually does something if invoked
 Sven> with zero or only one argument, for example).

Yes.  And since _call shall allow people to customize the completion system
for those cases, it's not a problem anyway.  

[...]

 Sven> So the style would only be used to get the command (including: how it
 Sven> should be invoked, i.e. with `command' or not) and any options the
 Sven> user wants to give to it. If the style is not set we use some standard 
 Sven> way, so we don't have to set up default styles for this.

This (your last sentence) assume that no option is used by default (like
the call to ps in _pids).  This seems restrictive (but I don't have an
example where options are needed by default, and where the user would want
to change them).

 >> ...
 >> 
 >> Another point about the $+functions[<name>] test: what if I am writting a
 >> completion function for a shell function? say I need to call it, how do I
 >> do?

 Sven> Good point. Also testing $+commands and $+builtins might help here,
 Sven> but could still be wrong. Hm, I just wanted to make this cleverer but
 Sven> since the style would allow one to override it anyway, we should
 Sven> probably just call it without any pre-command modifier in the default
 Sven> case. Or let _call accept options like -c and -b to say that the
 Sven> default should use `command' or `builtin'. 

Given I am writting a _call in a completion function, how do I decide
whether I must use -[bc] or not?  We ne a rule here, that should be used
consistently in the completion system (see the first paragraph for what I
suggest). 

And if a rule is chosen, _call can apply it, and therefore -[bc] parameters
may not be needed anymore.

[...]

-- 
Alexandre Duret-Lutz