On 20/04/2024 23.49, Andrew Tomazos wrote:
> Jens,
> **
> 16.5% of if statements are like `if (!expr)`

That wasn't my question.

For plain if (!expr), I see absolutely zero
benefit in special-casing the ! and moving it outside
of the parens: if ! (expr)
It saves zero tokens and introduces more syntax.

The claim was that we can sometimes save a level of nested
parentheses:

  if (! (a && b) )

becomes

  if ! (a && b)

The numbers given by the author were originally referring
to this case, and, given those numbers, I'm not seeing this
as a frequent enough situation to warrant the extra syntax.


Note that

  if (! (a && b) || c)

can't become

  if ! (a && b || c)


Jens

 
> You're welcome.
> -Andrew.
>
> PS (`if(`=7495313 / `if(!`=45254716) in ACTCD19

These numbers feel swapped.

> PSS Yes, I checked `if(!` was dominated by `if(!expr)`




> On Sun, Apr 21, 2024 at 7:05 AM Jens Maurer via Std-Proposals <std-proposals_at_[hidden] <mailto:std-proposals_at_[hidden]>> wrote:
>
>
>
> On 20/04/2024 21.34, Yexuan Xiao wrote:
> > Thank you for pointing out the error in my proposal; I’ll try to correct it in a later version.
> >
> > The data from LLVM or GCC shows code that directly benefits from this. Additionally, I searched all the Ruby code in Rails (a Ruby web framework with many users), where ‘unless’ appeared 1808 times, while ‘if’ appeared 10157 times. This data demonstrates the potential of the ‘if not’ pattern.
>
> You're answering a question I didn't ask.
>
> I asked: How many "if" statements are there in LLVM in total?
>
> By my crude count, for all of "llvm-project":
>
> $ find . -name *.h -o -name *.cpp | xargs grep -w '[^#]if' | wc -l
> 498710
>
> That's far less than 1% of "if" statements that would benefit.
> I'm not convinced by those numbers.
>
> > C++23 accepted ‘if !consteval’ instead of ‘if consteval(false)’, and the previous proposal P1181 has revealed other advantages, so I believe adding ‘if !constexpr’ or simply ‘if !’ is not a wart.
>
> You're welcome to disagree with me.
>
> The ISO standardization process is a consensus-driven process,
> and I'm herewith advising you, as early in the process as possible,
> of my opinion regarding your proposal.
>
> Absent further convincing rationale, I believe your proposal does
> not provide sufficient benefits in return for the added complexity,
> and I'll vote strongly against your proposal at any opportunity
> where I have a chance to do so.
>
> I hope that clarifies my position.
>
> Jens
>
>
> ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
> > *From:* Jens Maurer <jens.maurer_at_[hidden] <mailto:jens.maurer_at_[hidden]>>
> > *Sent:* Sunday, April 21, 2024 3:01
> > *To:* std-proposals_at_[hidden] <mailto:std-proposals_at_[hidden]> <std-proposals_at_[hidden] <mailto:std-proposals_at_[hidden]>>
> > *Cc:* Yexuan Xiao <bizwen_at_[hidden] <mailto:bizwen_at_[hidden]>>
> > *Subject:* Re: [std-proposals] if !(condition)
> >
> >
> >
> > On 20/04/2024 18.09, Yexuan Xiao via Std-Proposals wrote:
> >> I've wrote a proposal that suggests allowing the omission of the outermost parentheses to simplify the conditions in if statements:
> >> https://aus01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fstorage.nykz.org%2Fproposals%2Fif-not%2F&data=05%7C02%7C%7Cb53d7d71b9eb4c67433f08dc616c49c7%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C638492364927500272%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=Njw2Jwu1tWH8azBLMKGy6%2F2GPKqCr2uYCFBoxqsdvDc%3D&reserved=0 <https://aus01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fstorage.nykz.org%2Fproposals%2Fif-not%2F&data=05%7C02%7C%7Cb53d7d71b9eb4c67433f08dc616c49c7%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C638492364927500272%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=Njw2Jwu1tWH8azBLMKGy6%2F2GPKqCr2uYCFBoxqsdvDc%3D&reserved=0> <https://storage.nykz.org/proposals/if-not/ <https://storage.nykz.org/proposals/if-not/>>
> >>
> >> Please share your comments.
> >
> > A "condition" might be a declaration. The "equivalent" syntax
> > "!(condition)" is not valid C++ if the "condition" is a declaration.
> >
> > Thus, your proposal seems incomplete.
> >
> > You quote 1500 simplification opportunities in LLVM and 2300 ones
> > in GCC. How many "if" statements are there in total, in each of
> > those codebases?
> >
> > Beyond that, I'm strongly opposed: this delivers no benefit at all.
> > Being able to omit nested parentheses in the condition in some
> > cases (but not others, if the negation isn't at the top level)
> > is insufficient motivation for another wart in the language,
> > in my view.
> >
> > Someone hiding expression syntax in a macro is just misguided;
> > I have no sympathy whatsoever.
> >
> > Jens
> >
> --
> Std-Proposals mailing list
> Std-Proposals_at_[hidden] <mailto:Std-Proposals_at_[hidden]>
> https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals <https://lists.isocpp.org/mailman/listinfo.cgi/std-proposals>
>