Re: [PATCH 0/2] grep: introduce and use grep_and_expr()

["René Scharfe" <l.s.r@web.de>]

Am 07.01.22 um 20:33 schrieb Taylor Blau:
> On Fri, Jan 07, 2022 at 01:57:17PM +0100, René Scharfe wrote:
>> Am 06.01.22 um 20:50 schrieb Taylor Blau:
>>> René,
>>>
>>> Here are a couple of extra patches on top of your series which introduce
>>> and use a new grep_and_expr() function.
>>>
>>> Like the final patch says, this isn't about reducing code duplication,
>>> but rather about adding visual consistency with the other
>>> `compile_pattern_xyz()` functions.
>>>
>>> Taylor Blau (2):
>>>   grep: extract grep_binexp() from grep_or_expr()
>>
>> I considered extracting such a function as well.  I'd have called it
>> grep_binary_expr(), though, to match the existing names.
>>
>> I decided against it because it can be misused by passing a non-binary
>> kind to it.  (That's a weak objection, but the benefit of such a
>> function was low already in my mind because it doesn't do much.)  You
>> solve this by keeping grep_or_expr() and adding grep_and_expr(), which
>> cannot be misused in this way -- OK.

That "OK" should probably have been a "Good".

>
> That makes sense. If it's keeping you up at night, we could easily add a
> check to ensure that `kind` is one of GREP_NODE_OR or GREP_NODE_AND. But
> I think that any new code that looks like:
>
>     grep_binexp(GREP_NODE_NOT, xyz, NULL);
>
> would probably stick out like a sore thumb. So I doubt that such a check
> would buy us much practically speaking.

Having the wrappers for all the binary node types is good enough.  I'm
not so sure about the sticking out thing -- I managed to miss bigger
mistakes before..

> But I agree that this whole thing probably isn't worth the minimal
> effort required, since the couple of patches I posted on top are purely
> about cosmetics.
>
> All of that is to say that I'd be happy to see these patches picked up,
> and I would also not be sad at all to see them left on the floor.

I think the end result is an improvement.

René