Thinking about this a little more, I'm now attracted to the idea that
its .gitignore that's weird.
As I understand it, .gitignore stops recursion when there's a
directory match (`somedir/`) but also explicitly allows nested
.gitnore file _as well as_ exclusion (`!*.txt`).
So, in the following (contrived) example, the user doesn't get what they want:
repo/
|- .git/
|- .gitignore # /ignore-most/
|- ignore-most/
| |- .gitignore # !*.txt
| |- please_ignore.png
| |- dont_ignore_me.txt
`repo/ignore-most/dont_ignore_me.txt` is still ignored, despite what
seems like the obvious intention of the user.
Maybe a unified "best-practices" would first-and-foremost recommend
against matching directories at all (makes sense, git doesn't manage
directories). In the above example, changing `/ignore-most/` to
`/ignore-most/*` has the "desired" effect.
What do you think?
On Tue, Mar 20, 2018 at 12:28 PM, Duy Nguyen <pclouds@xxxxxxxxx> wrote:
> On Tue, Mar 20, 2018 at 5:40 AM, Jeff King <peff@xxxxxxxx> wrote:
>> On Tue, Mar 20, 2018 at 12:25:27AM -0400, Dakota Hawkins wrote:
>>
>>> > Right. The technical reason is mostly "that is not how it was designed,
>>> > and it would possibly break some corner cases if we switched it now".
>>>
>>> I'm just spitballing here, but do you guys think there's any subset of
>>> the combined .gitignore and .gitattributes matching functionality that
>>> could at least serve as a good "best-practices, going forward"
>>> (because of consistency) for both? I will say every time I do this for
>>> a new repo and have to do something even slightly complicated or
>>> different from what I've done before with .gitattributes/.gitignore
>>> that it takes me a long-ish time to figure it out. It's like I'm
>>> vaguely aware of pitfalls I've encountered in the past in certain
>>> areas but don't remember exactly what they are, so I consult the docs,
>>> which are (in sum) confusing and lead to more time spent
>>> trying/failing/trying/works/fails-later/etc.
>>>
>>> One "this subset of rules will work for both this way" would be
>
> You know, you (Dakota) could implement the new "exclude" attribute in
> .gitattributes and ignore .gitignore files completely. That makes it
> works "for both" ;-) The effort is probably not small though.
>
>>> awesome even if the matching capabilities are technically divergent,
>>> but on the other hand that might paint both into a corner in terms of
>>> functionality.
>>
>> As far as I know, they should be the same with the exception of this
>> recursion, and the negative-pattern thing. But I'm cc-ing Duy, who is
>> the resident expert on ignore and attributes matching (whether he wants
>> to be or not ;) ).
>
> Ha ha ha.
>
>> I wouldn't be surprised if there's something I don't know about.
>
> The only thing from the top of my head is what made me fail to unify
> the implementation of the two. It's basically different order of
> evaluation [1] when your patterns are spread out in multiple files. I
> think it makes gitattr and gitignore behavior different too (but I
> didn't try to verify).
>
> Apart from that, the two should behave the same way besides the
> exceptions you pointed out.
>
> [1] https://public-inbox.org/git/%3CCACsJy8B8kYU7bkD8SiK354z4u=sY3hHbe4JVwNT_1pxod1cqUw@xxxxxxxxxxxxxx%3E/
>
>> So I think the "recommended subset" is basically "everything but these
>> few constructs". We just need to document them. ;)
>>
>> I probably should cc'd Duy on the documentation patch, too:
>>
>> https://public-inbox.org/git/20180320041454.GA15213@xxxxxxxxxxxxxxxxxxxxx/
>>
>> -Peff
> --
> Duy
