Re: Updates to the Plugin Standard

[Sebastian Gniazdowski <sgniazdowski@xxxxxxxxx>]

pt., 25 paź 2019, 21:08 użytkownik <gi1242+zsh@xxxxxxxxx> napisał:

> I took a quick look at the namespace prefix section: That looks a bit
> overkill to me. Also using . and / as function prefixes can easily get
> confused with paths.
>
> I'd suggest using just one name space prefix for everything to keep
> things simple. It could be ::function, or pluginName::funciton for
> instance.
>

The section has purely advisory character. However for the pluginName::
proposal there's a valid argument - the possible conflict as in
https://github.com/zdharma/fast-syntax-highlighting/issues/157.

As for the points about the dot and / prefixes, I'll take them into
consideration. Right now I can tell that the / prefix is used for debug
functions because they're most outside from the main code. They're side to
the main code, hence the overall nice (and most of all – available) but
otherwise problematic ASCII character '/' has been selected to them. The
other available characters, like ^ aren't available because of the <= 5.7.1
autoload bug (https://www.zsh.org/mla/workers/2019/msg00534.html) however
hence / isn't available for autoload anyway, the character might get
changed. Any propositions for a replacement?

If there are too many types, it makes it more confusing / cumbersome
> when writing.
>

My subjective impression is that the : prefix is especially useful. Many
functions are hooks, and it's nice to list them as in
https://asciinema.org/a/277059, just to see them grouped (noticing: they're
cannot be directly completed, entering :<TAB> tries to complete the
modifiers, is there a way to change this?).

I foresee that the number of prefixes will not get smaller, it might get,
however, larger, when the Unicode characters will get mentally accepted
into consideration. I'm already using …-prefix in one of my projects as
it's easy to type (Alt-k on my keymap), to denote a "backend function",
i.e. a function that is a closed black-box -like thing that is getting the
requests to do the dirty work at the end of a multi-stage preparement. The
function is called: …zp-test-make-targets.

GI
>
> --
> 'Common' Proof Techniques:
> 12. Proof by obfuscation -- A long plotless sequence of true and/or
> meaningless syntactically related statements.
>

>