On Fri, May 11, 2018 at 5:05 PM, SZEDER Gábor [off-list ref] wrote:

On Thu, May 10, 2018 at 10:46 AM, Nguyễn Thái Ngọc Duy
[off-list ref] wrote:

quoted

Instead of maintaining a separate list of command classification,
which often could go out of date, let's centralize the information
back in git.

While the function in git-completion.bash implies "list porcelain
commands", that's not exactly what it does. It gets all commands (aka
--list-cmds=main,others) then exclude certain non-porcelain ones. We
could almost recreate this list two lists list-mainporcelain and
others. The non-porcelain-but-included-anyway is added by the third
category list-complete.

list-complete does not recreate exactly the command list before this
patch though. The following commands are not part of neither
list-mainporcelain nor list-complete and as a result no longer
completes:

- annotate             obsolete, discouraged to use
- difftool-helper      not an end user command
- filter-branch        not often used
- get-tar-commit-id    not often used
- imap-send            not often used
- interpreter-trailers not for interactive use
- lost-found           obsolete
- p4                   too short and probably not often used (*)
- peek-remote          deprecated
- svn                  same category as p4 (*)
- tar-tree             obsolete
- verify-commit        not often used

'git name-rev' is plumbing as well.

So? name-rev remains completable like before and is not mentioned in
the above list. Am I missing something?

I think this commit should be split into two:

  - first do the unequivocally beneficial thing and get rid of the
    long, hard-coded command list in __git_list_porcelain_commands(),
    while keeping its output unchanged,

  - then do the arguable thing and change the list of commands.

I will. Though the first commit still changes the output slightly
because there are three commands not in command-list.txt. To keep the
output unchanged, I would need to add them back in command-list.txt
first (and find the right group for them) only to remove those lines
later (two of them deprecated, the other does not even have a man
page). It's not worth the effort.

quoted

 {
        local i IFS=" "$'\n'
-       for i in $(__git_commands)
+       for i in $(__git_commands $1)
        do
                case $i in
                *--*)             : helper pattern;;

Is this loop to exclude helper commands with doubledash in their name
still necessary?

It is needed for __git_list_all_commands() because that one still
essentially grabs "git help -a", which includes command--helpers.
-- 
Duy