[llvm-dev] [RFC] Coding Standards: "prefer `int` for regular arithmetic, use `unsigned` only for bitmask and when you intend to rely on wrapping behavior."

Tue Jun 11 01:01:25 PDT 2019

On Tue, Jun 11, 2019, 9:51 AM Mehdi AMINI <joker.eph at gmail.com> wrote:

>
>
> On Tue, Jun 11, 2019 at 12:37 AM Aaron Ballman <aaron.ballman at gmail.com>
> wrote:
>
>> Sorry for the brevity, I am currently travelling and responding on a cell
>> phone. I won't be able to give you a full accounting until later,
>>
>
> There is no hurry right now!
>
>
>> but 1) I don't see a motivating problem this churn solves, 2) signed int
>> does not represent the full size of an object like size_t does and is
>> inappropriate to use for addressing into objects or arrays, which means we
>> won't use this convention consistently anyway.
>>
>
> As far as I can tell, the same arguments applies to "unsigned" I believe:
> it does not represent the full extent of size_t on every platform either!
> Yet we're still using it as an index, including to index into objects/array.
>

>
>
>> I have yet to be convinced by the c++ community's very recent desire to
>> switch everything to signed integers and would be very unhappy to see us
>> switch without considerably more motivating rarionale.
>>
>
> Note that this is not about "switching" anything: there is no standard in
> LLVM right now (as far as I know) and the codebase is inconsistent (I am
> using `int` in general for a while I believe).
>

Glad to hear there's no plan to change the codebase, but then I fail to see
the purpose of the coding rule. Won't it boil down to "use the correct
datatype for what you're doing?"

What I mostly want to avoid is code that uses signed values for inherently
unsigned operations like comparing against the size of an object. Then we
start wanting ssize() functions or using casts for confused reasons.

~Aaron

> --
> Mehdi
>
>
>
>>
>> On Mon, Jun 10, 2019, 11:04 PM Mehdi AMINI <joker.eph at gmail.com> wrote:
>>
>>>
>>>
>>> On Mon, Jun 10, 2019 at 10:32 AM Aaron Ballman via llvm-dev <
>>> llvm-dev at lists.llvm.org> wrote:
>>>
>>>>
>>>>
>>>> On Mon, Jun 10, 2019, 7:16 PM Jake Ehrlich via llvm-dev <
>>>> llvm-dev at lists.llvm.org> wrote:
>>>>
>>>>> I'm in the same situation James is in and thus have the same bias but
>>>>> I'll +1 that comment nevertheless. I think I prefer using size_t or the
>>>>> uintX_t types where applicable. Only when I need a signed value do I use
>>>>> one.
>>>>>
>>>>
>>>> +1 to prefering unsigned types.
>>>>
>>>
>>> I'd appreciate if you guys could provide rational that address the
>>> extensive arguments and opinion provided in the C++ community that I tried
>>> to summarize in the link above.
>>> Otherwise I don't know what to take out of unmotivated "+1".
>>>
>>> --
>>> Mehdi
>>>
>>>
>>>
>>>>
>>>>> On Mon, Jun 10, 2019, 9:59 AM James Henderson via llvm-dev <
>>>>> llvm-dev at lists.llvm.org> wrote:
>>>>>
>>>>>> Maybe it's just because I work in code around the binary file formats
>>>>>> almost exclusively, but unsigned (or more often uint64_t) is FAR more
>>>>>> common than int everywhere I go. I don't have time right now to read up on
>>>>>> the different links you provided, and I expect this is covered in them, but
>>>>>> it also seems odd to me to use int in a loop when indexing in a container
>>>>>> (something that can't always be avoided), given the types of size() etc.
>>>>>>
>>>>>> On Mon, 10 Jun 2019 at 17:26, Michael Kruse via llvm-dev <
>>>>>> llvm-dev at lists.llvm.org> wrote:
>>>>>>
>>>>>>> Am Sa., 8. Juni 2019 um 13:12 Uhr schrieb Tim Northover via llvm-dev
>>>>>>> <llvm-dev at lists.llvm.org>:
>>>>>>> > I'd prefer us to have something neater than static_cast<int> for
>>>>>>> the
>>>>>>> > loop problem before we made that change. Perhaps add an ssize (or
>>>>>>> > equivalent) method to all of our internal data structures? They're
>>>>>>> a
>>>>>>> > lot more common than std::* containers.
>>>>>>>
>>>>>>> +1
>>>>>>>
>>>>>>> Since C++20 is also introducing ssize [1] members, this makes a lot
>>>>>>> of
>>>>>>> sense to me. Using it would help avoiding an unsigned comparison as
>>>>>>> in
>>>>>>>
>>>>>>>     if (IndexOfInterestingElement >= Container.size())
>>>>>>>       ...
>>>>>>>
>>>>>>> to sneak in from the start.
>>>>>>>
>>>>>>> Michael
>>>>>>>
>>>>>>> [1] http://wg21.link/p1227r1
>>>>>>> _______________________________________________
>>>>>>> LLVM Developers mailing list
>>>>>>> llvm-dev at lists.llvm.org
>>>>>>> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>>>>>>>
>>>>>> _______________________________________________
>>>>>> LLVM Developers mailing list
>>>>>> llvm-dev at lists.llvm.org
>>>>>> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>>>>>>
>>>>> _______________________________________________
>>>>> LLVM Developers mailing list
>>>>> llvm-dev at lists.llvm.org
>>>>> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>>>>>
>>>> _______________________________________________
>>>> LLVM Developers mailing list
>>>> llvm-dev at lists.llvm.org
>>>> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>>>>
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20190611/0816b656/attachment-0001.html>