Masatake YAMATO <[EMAIL PROTECTED]> writes:

>> > Probably a good general convention would be
>> >
>> > * Capital letters => prefix for other commands
>> >
>> > * Lower-case letters => commands.
>> >
>> > That's mainly how Gnus works.
>> 
>> That is a good strategy.
>
> Do you think we should apply this strategy to all the other bindings?
No - see below.
But I think we should use that convention, when we add new
functionality and run out of keys.
> We have already defined many key bindigs which break the new strategy.
>
>     (defvar tla--key-help        ??)        ; help
>     (defvar tla--key-mark-prefix ?*)   ; other mark related command prefix
>     (defvar tla--key-apply-from-here ?.)    ; apply something from here
>     (defvar tla--key-get      ?>)           ; prefix for getting something
>     (defvar tla--key-mirror   ?<)           ; mirror
>     (defvar tla--key-parent   ?^)       ; visit uppper XXX. e.g. directory
>     (defvar tla--key-diff     ?=)           ; one shot

Let's keep them.

>     (defvar tla--key-add      ?a)           ; prefix for adding something
>     (defvar tla--key-diff-prefix ?d)
>     (defvar tla--key-remove   ?r)           ; prefix for remove something
>     (defvar tla--key-toggle   ?t)           ; prefix for toggle
>     (defvar tla--key-kill-ring-prefix ?w)
We should include the -prefix suffix for that variables.
I would keep them also, because they are easier to type

>     (defvar tla--key-ediff    ?e)
>     (defvar tla--key-refresh  ?g)           ; update or refresh buffer
>     (defvar tla--key-mark     ?m)           ; mark
>     (defvar tla--key-next     ?n)           ; next item
>     (defvar tla--key-previous ?p)           ; previous item
>     (defvar tla--key-quit     ?q)           ; quit
>     (defvar tla--key-unmark   ?u)           ; unmark
That are no prefix keys - are they?

>     (defvar tla--key-buffer-prefix ?B)   ; perfix for switching XXX buffer
>     (defvar tla--key-directory-prefix ?D)
>     (defvar tla--key-merge-prefix ?M)
>     (defvar tla--key-revert ?U)
>     (defvar tla--key-working-copy ?W)       ; Affecting on working copy

These follow the new (additional) guidelines
We can keep or revert the ?U binding
We should rename tla--key-working-copy to tla--key-working-copy-prefix

> Also we have following rules and conflicts with the new strategy.
>
> ;; 3. Upper case for commands taking longer time to be executed.
> ;; 4. Lowwer case for commands taking shorter time to be executed.

I consider that rule as lower priority.
It would apply for commands like ?W ?c, ?W ?C

-- 
Stefan.

Reply via email to