Complete global aliases?

Complete global aliases?

[Sebastian Gniazdowski]

[-- Attachment #1: Type: text/plain, Size: 367 bytes --]

Hi,
I think that the global aliases should be completed out of the box, what do
you say? In a project I've created a bunch of global aliases to allow
comfortable jumpings to its directories to become disappointed that they
(global aliases) aren't completed like the named directories are. IMO this
should be fixed on the upstream. Could someone add a snippet for it?

[-- Attachment #2: Type: text/html, Size: 419 bytes --]

Re: Complete global aliases?

[Bart Schaefer]

On Sat, Sep 17, 2022 at 7:27 AM Sebastian Gniazdowski
<sgniazdowski@gmail.com> wrote:
>
> I think that the global aliases should be completed out of the box, what do you say?

I think there needs to be a bit more specification about how this would work.

Where would this fit in the overall design of contexts?  E.g. does it
happen in "-default-" context (which isn't always applied) and if so
do alaises complete instead of files, as well as files, or only when
there are no files (that is, _files returns nonzero)?

Do the aliases expand or is just their name completed?  What happens
if you complete to the right of a global alias that (if expanded)
changes the syntax of the command line?

There are probably several other cases to consider, depending on the
answers to those.

Re: Complete global aliases?

[Eric Cook]

On 9/17/22 10:20, Sebastian Gniazdowski wrote:
> Hi,
> I think that the global aliases should be completed out of the box, what do you say? In a project I've created a bunch of global aliases to allow comfortable jumpings to its directories to become disappointed that they (global aliases) aren't completed
> like the named directories are. IMO this should be fixed on the upstream. Could someone add a snippet for it?

There are dynamic named directories for this very thing.

Re: Complete global aliases?

[Sebastian Gniazdowski]

[-- Attachment #1: Type: text/plain, Size: 1075 bytes --]

On Sun, 18 Sept 2022 at 01:27, Bart Schaefer <schaefer@brasslantern.com>
wrote:

> On Sat, Sep 17, 2022 at 7:27 AM Sebastian Gniazdowski
> <sgniazdowski@gmail.com> wrote:
> >
> > I think that the global aliases should be completed out of the box, what
> do you say?
>
> I think there needs to be a bit more specification about how this would
> work.
>
> Where would this fit in the overall design of contexts?  E.g. does it
> happen in "-default-" context (which isn't always applied) and if so
> do alaises complete instead of files, as well as files, or only when
> there are no files (that is, _files returns nonzero)?
>

I'm not sure, however there's a solution on the web, see:
https://stackoverflow.com/a/59513051/11999858

>
> Do the aliases expand or is just their name completed?


I think that only its name should be completed.



> What happens
> if you complete to the right of a global alias that (if expanded)
> changes the syntax of the command line


I think  that it shouldn't be expanded, so the issue isn't actual.
-- 
Best regards,
Sebastian Gniazdowski

[-- Attachment #2: Type: text/html, Size: 2724 bytes --]

Re: Complete global aliases?

[Bart Schaefer]

On Sun, Sep 18, 2022 at 2:43 AM Sebastian Gniazdowski
<sgniazdowski@gmail.com> wrote:
>
> On Sun, 18 Sept 2022 at 01:27, Bart Schaefer <schaefer@brasslantern.com> wrote:
>>
>> Where would this fit in the overall design of contexts?
>
> I'm not sure, however there's a solution on the web, see: https://stackoverflow.com/a/59513051/11999858

That solution gives the user control over when global aliases are
completed, by adding a function that can be placed anywhere in the
completer zstyle.  The suggested value of that zstyle happens to make
global aliases complete before anything else, but that doesn't answer
the question of how a default should work.

>> Do the aliases expand or is just their name completed?
>
> I think that only its name should be completed.

I'll note that theoretically the ^Xa key binding (expand aliases) can
be programmed to complete when a partial word matches an alias and
expand when a full word matches.  I didn't attempt setting up the
necessary zstyle contexts.

>> What happens
>> if you complete to the right of a global alias that (if expanded)
>> changes the syntax of the command line
>
> I think  that it shouldn't be expanded, so the issue isn't actual.

Upon further thought, I agree that it's not an issue for completion of
an alias itself.  More generally, though, if the expansion of a global
alias contains ";" or "|" or "&&" or "||" etc., then ideally further
completions shouldn't be using the context of the preceding command
word.  Unfortunately that requires checking every word to the left of
the cursor for a global alias, and expanding it, to get the "real"
context.

Re: Complete global aliases?

[Sebastian Gniazdowski]

[-- Attachment #1: Type: text/plain, Size: 2291 bytes --]

On Mon, 19 Sept 2022 at 03:46, Bart Schaefer <schaefer@brasslantern.com>
wrote:

> On Sun, Sep 18, 2022 at 2:43 AM Sebastian Gniazdowski
> <sgniazdowski@gmail.com> wrote:
> >
> > On Sun, 18 Sept 2022 at 01:27, Bart Schaefer <schaefer@brasslantern.com>
> wrote:
> >>
> >> Where would this fit in the overall design of contexts?
> >
> > I'm not sure, however there's a solution on the web, see:
> https://stackoverflow.com/a/59513051/11999858
>
> That solution gives the user control over when global aliases are
> completed, by adding a function that can be placed anywhere in the
> completer zstyle.  The suggested value of that zstyle happens to make
> global aliases complete before anything else, but that doesn't answer
> the question of how a default should work.
>
Maybe looking from named dirs perspective can help? Currently, cd ~HO<tab>
doesn't complete, too. I think that lack of such an obvious feature can
make forward a way of completing also global aliases (like alias -g HOME
~)? How should be ~HOME completed I think that everybody knows (with
exception of me)? The global aliases should be completed the same.


> >> Do the aliases expand or is just their name completed?
> >
> > I think that only its name should be completed.
>
> I'll note that theoretically the ^Xa key binding (expand aliases) can
> be programmed to complete when a partial word matches an alias and
> expand when a full word matches.  I didn't attempt setting up the
> necessary zstyle contexts.
>
> When I tried ^Xa on my incomplete global alias it didn't complete…


> > I think  that it shouldn't be expanded, so the issue isn't actual.
>
> Upon further thought, I agree that it's not an issue for completion of
> an alias itself.  More generally, though, if the expansion of a global
> alias contains ";" or "|" or "&&" or "||" etc., then ideally further
> completions shouldn't be using the context of the preceding command
> word.  Unfortunately that requires checking every word to the left of
> the cursor for a global alias, and expanding it, to get the "real"
> context.
>

Could it just complete like ~HO<tab> should ? I don't really see the
context issue, HO and ~HO should just complete…

-- 
Best regards,
Sebastian Gniazdowski

[-- Attachment #2: Type: text/html, Size: 3904 bytes --]