Re: Any exceptions for the 15-line rule?

[Dmitry Gutov <dgutov@yandex.ru>]

On 27.04.2013 17:28, Stefan Monnier wrote:
>> Thanks, `substring' is better than `replace-match' I mentioned. But
>> still, should this be considered a full new implementation? Does
>> replacing `cond' with `if' in the inner condition make it a new piece of
>> code, as opposed to derivative one?
>
> The purpose is just to "simplify" the code, rather than to obscure
> the copyright.  In terms of copyright, it does reduce the amount of code
> taken, indeed, but it's not a very significant difference.

I see.

>>> With such cleanups, the patch seems acceptable as a "tiny change".
>>> But please do ask for the CA as well (so the use of "tiny change" is
>>> mostly a way to avoid having to wait for the CA to go through).
>> To be clear, who do I ask to sign the CA over the modified patch? The
>> auto-complete-clang author, or the person who looked at a few pieces
>> from that package and adapted them to (admittedly, fairly similar)
>> company-clang code?
>
> In terms of who owns the copyright, the answer is probably "both", but
> to the extent that it fits the "tiny change" criteria we don't need to
> care too much (unless one or both of the authors already have
> contributed code as a "tiny change" since those things are cumulative).

> Assuming that we want company-mode and auto-complete to share more code
> in the future, having the assignment of AC's author is a good idea.

auto-complete-clang is a separate package from auto-complete (which has 
multiple authors, but the principal one already has CA on file: Tomohiro 
Matsuyama). I'll ask, although I think the only useful piece left we 
could borrow is converting the lightweight markup in completion 
candidates into syntax highlighting, and that I should be able to write 
in some different way without much difficulty.

> As for the person who sent you the patch, it would also make sense to
> get his/her assignment if there's a chance he'll contribute more in
> the future.

Ok, I'll ask.

> PS: By the way, I think company-backends should be merged with (and/or
> moved over to) completion-at-point-functions, and some of those backends
> should be moved to their respective major modes.

Yes, I remember the email that accompanied your company-capf patch, but 
so far I don't see a good way to go about it.

The interface provided by company-backends is considerably richer. How 
do we fit that inside completion-at-point-functions?

There's also the problem of keeping compatibility with already released 
Emacs versions. If we move the code, they won't be able to use it; if we 
keep it, it would create duplication.