Re: [elpa] master 24746ff 1/8: Drop Names dependency

[Artur Malabarba <bruce.connor.am@gmail.com>]

On 13 Oct 2015 8:22 pm, "Stefan Monnier" <monnier@iro.umontreal.ca> wrote:
>
> >     Drop Names dependency
>
> Curious: why?
>
> PS: I'd guess that you ended up preferring nameless.el, but I'm
> interested in the actual details of why/how.

Yes I'm currently preferring nameless, though it's just a first assessment.

Comparing the two, they perform very similarly. Names is slightly
nicer when typing [1] and Nameless is slightly nicer when reading [2].
The difference that tips the scale is that Names comes with some
luggage.

- It imposes dependencies at autoload time.
- It makes completion a little less helpful (company-dabbrev
completion still works, but company-capf doesn't).
- It makes code-navigation a little more difficult (it does some
hacking so that things like `find-function/variable' still work, which
is all I ever use, but there are other methods and they will fail)
- It is a proper package dependency that gets installed on the user's
computer. This isn't really a failure. It's just slightly less
convenient than Nameless which is not a dependency.

Since Nameless is just completely clean, I'm leaning more towards it
right now. Not that I'm any less proud of Names. :-)
Sometimes you gotta climb a mountain to see that there's another way.


Artur

[1]: Not having to type the namespace is better than having a key that
inserts the namespace.
[2]: Both hide the namespace, but Nameless adds a colored ":" to the
symbol, which differentiates locals from globals and makes the code
slightly easier to understand.