Improving completion of subcommands

[NightMachinery]

The service variable gets used by, e.g., git. Not setting it will break those completions.

[AdrieanKhisbe]

Thanks @NightMachinary, seems legit,

service was already mentioned in the OMZ issue I mentioned just below the line you injected it.

[AdrieanKhisbe]

@NightMachinary Two questions in order to have it land:

• would you mind I push to your branch to add tests
• if you have some link/docs that you saw to fix your issue for future reference

:)

[NightMachinery]

@AdrieanKhisbe commented on Aug 23, 2020, 11:57 PM GMT+4:30:

@NightMachinary Two questions in order to have it land:

• would you mind I push to your branch to add tests
• if you have some link/docs that you saw to fix your issue for future reference

:)

I don't mind at all. Should I do sth for this to be possible? (Should I add you as a contributor to my fork?)

I figured this out by examining the completion function _git, so no docs. I have been using it since then, it's pretty good. which works with this, but it didn't work before. I also saw it in man zshall when I was looking for a more robust way to find completion functions (using _name is quite unoptimal. I myself used that convention for my helper functions, and the completion used, e.g., my _mv (which actually moves files and has nothing to do with completions) when trying to complete mv. I got rid of those in my own code, but I guess the problem still exists in the wild. You might want to add a warning to the readme.):

Each name may also be of the form `cmd=service'.  When completing the command cmd, the function typ-
              ically  behaves  as  if  the command (or special context) service was being completed instead.  This
              provides a way of altering the behaviour of functions that can perform many  different  completions.
              It  is  implemented  by  setting  the parameter $service when calling the function; the function may
              choose to interpret this how it wishes, and simpler functions will probably ignore it.

[AdrieanKhisbe]

About the branch, normally no intervention need, (discovered it by myself few months ago by accident).

Thanks for the details,I found the warning you just suggested in the zsh completion doc

Will try to take care of it tomorrow or tuesday.
(if you have no news by wednesday, feel free to ping me here)

[AdrieanKhisbe]

@NightMachinary I think I as something see gh actions,
and it's a good thing I added test: I spotted that we need another $service injection to have both mybookmark git something (whitelisted command) and mybookmark - git something (any command).

Would you mind to have a look before I merge this?

[NightMachinery]

@AdrieanKhisbe commented on Aug 24, 2020, 4:16 PM GMT+4:30:

@NightMachinary I think I as something see gh actions,
and it's a good thing I added test: I spotted that we need another $service injection to have both mybookmark git something (whitelisted command) and mybookmark - git something (any command).

Would you mind to have a look before I merge this?

I took a look. Indeed, your tests were a good idea! I am using my own minimal clone of diractions, which has alternative design choices, so I did not catch that in my usage.

[AdrieanKhisbe]

@NightMachinary et voilà, it's merged.

Thanks for your contribution 🙂