Re: [ELPA] New package: transient

[조성빈]

> 2020. 5. 3. 오전 3:46, Eli Zaretskii <address@hidden> 작성:
> 
> 
>> From: 조성빈 <address@hidden>
>> Date: Sun, 3 May 2020 03:37:31 +0900
>> Cc: Stefan Monnier <address@hidden>,
>> address@hidden,
>> address@hidden,
>> address@hidden
>> Eli Zaretskii <address@hidden> 작성:
>>>> My understanding is that slow here means that opening a new *Apropos*
>>>> buffer is an small, but additional mental burden when writing code, and
>>>> it slows down writing code.
>>> How is that different from having *Completions* pop up instead?
>> Which… is why a lot of people don’t use *Completion* and use company-mode
> 
> Which does the same, just in a menu.  How's that different?
> 
>>> And how is looking up the function you need a "burden", when any
>>> modern IDE provides some way of showing the possible candidates for
>>> what you want to do next?
>> Modern IDEs provide the candidates without needing to do any action.
> 
> How's that relevant to the issue at hand?  It's a tangent.
> 
>> It’s very different from having to explicitly look up.
> 
> No, it isn't.  "Look up" means look through the list of candidates,
> applying some logic towards the decision which candidate to choose.

I’m assuming you aren’t using auto completion here — sorry if it’s not true.

Looking up with C-h means one has to explicitly trigger a search with a search 
term. In contrast, with auto completion it’s basically searching with the 
latest token one is typing. Which means it has much less burden.

One of the many reasons I prefer the prefix scheme is because it works with 
completion well — which is probably also related to discoverability.

>>>> The regular naming scheme will mean that we can only search functions that
>>>> start with regexp - since the searcher doesn’t need grep-regexp-alist or
>>>> gmm-regexp-concat when trying to get regexp APIs.
>>> Do they need rx, as just one example?
>> I think that’s a different question.
>> (FWIW, I consider that rx shouldn’t appear.)
> 
> Let's see how the re- renaming thread evolves, I'm not sure everyone
> will agree.

Yes, that kind of bike shedding is the reason why this mailing list exists, 
right?
It’s great to me that there’s at least discussion about this here.

>> Whether or not the searcher wants to know about rx or not,
>> it’s probably true that one doesn’t want gmm-regexp-concat.
> 
> Why not?  Sounds like a very useful function for someone who works
> with regular expressions.

Oops, looks like I should really sleep. I was meaning grep-regexp-alist.