public inbox for libstdc++@gcc.gnu.org
 help / color / mirror / Atom feed
From: unlvsur unlvsur <unlvsur@live.com>
To: Jonathan Wakely <jwakely@redhat.com>
Cc: unlvsur unlvsur via Libstdc++ <libstdc++@gcc.gnu.org>
Subject: RE: Is it possible not to use snprintf or any other stdio functionalities to throw std::logic_error in standard library containers?
Date: Thu, 20 May 2021 19:47:13 +0000	[thread overview]
Message-ID: <DM6PR05MB469776570EE332423E1BE7F9D62A9@DM6PR05MB4697.namprd05.prod.outlook.com> (raw)
In-Reply-To: <YKa56hf4H+b0XEzo@redhat.com>

// I have disabled libstdcxx verbose

#include<fast_io.h>

int main(int argc,char** argv)
{
                std::string_view str(*argv);
                puts(str.substr(4).data());
}

g++ -o b b.cc -Ofast -std=c++20 -s -flto -march=native -I../fast_io/include -static-libstdc++

binary size 86kb


#include<fast_io.h>

int main(int argc,char** argv)
{
                std::string_view str(*argv);
                if(str.size()<4)
                                __builtin_trap();
                puts(std::string_view(str.data()+4,str.size()).data());
}

g++ -o c c.cc -Ofast -std=c++20 -s -flto -march=native -I../fast_io/include -static-libstdc++

binary size 15kb



Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10

From: Jonathan Wakely<mailto:jwakely@redhat.com>
Sent: Thursday, May 20, 2021 15:35
To: unlvsur unlvsur<mailto:unlvsur@live.com>
Cc: unlvsur unlvsur via Libstdc++<mailto:libstdc++@gcc.gnu.org>
Subject: Re: Is it possible not to use snprintf or any other stdio functionalities to throw std::logic_error in standard library containers?

On 20/05/21 19:27 +0000, unlvsur unlvsur wrote:
>Well. It comes from snprintf. You cannot use any stdio functionalities.
>
>For example
>https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fgcc-mirror%2Fgcc%2Fblob%2F4f4a2f199baf46d35492edadc16f30f32920c4df%2Flibstdc%252B%252B-v3%2Finclude%2Fbits%2Fstl_vector.h%23L1070&amp;data=04%7C01%7C%7C5e592cf75b1a4524c05408d91bc66216%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637571361118589300%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=5OZqIYtKRzXDWOcATMsIrBBYjCJOupkxk0of7SagS%2FQ%3D&amp;reserved=0

But that doesn't use snprintf.

>
>https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fgcc-mirror%2Fgcc%2Fblob%2F16e2427f50c208dfe07d07f18009969502c25dc8%2Flibstdc%252B%252B-v3%2Finclude%2Fexperimental%2Fstring_view%23L274&amp;data=04%7C01%7C%7C5e592cf75b1a4524c05408d91bc66216%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637571361118589300%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=iukICKyYF0pB1qJv16vCcZpXrbpBhzK%2FPteRZAEZx3U%3D&amp;reserved=0

And that doesn't use snprintf either.

Have you actually seen a dependency on stdio from these functions, or
are you just assuming?

>We need something which is a silent EH thrown, this format string thrown stuff will always introduce dependencies silently.
>
>This also hurts optimizations
>
>[cid:image002.png@01D74D8C.9AD99570]
>
>
>Sent from Mail<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgo.microsoft.com%2Ffwlink%2F%3FLinkId%3D550986&amp;data=04%7C01%7C%7C5e592cf75b1a4524c05408d91bc66216%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637571361118589300%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&amp;sdata=e5r3X770y%2Bm4nLrY37xW%2FUv3Bg7aa0bpsQEMoLnMstA%3D&amp;reserved=0> for Windows 10
>
>From: Jonathan Wakely<mailto:jwakely@redhat.com>
>Sent: Thursday, May 20, 2021 15:21
>To: Jonathan Wakely<mailto:jwakely.gcc@gmail.com>
>Cc: unlvsur unlvsur<mailto:unlvsur@live.com>; unlvsur unlvsur via Libstdc++<mailto:libstdc++@gcc.gnu.org>
>Subject: Re: Is it possible not to use snprintf or any other stdio functionalities to throw std::logic_error in standard library containers?
>
>On 15/05/21 09:23 +0100, Jonathan Wakely via Libstdc++ wrote:
>>On Sat, 15 May 2021, 04:12 unlvsur unlvsur via Libstdc++, <
>>libstdc++@gcc.gnu.org> wrote:
>>
>>> The issue is that this will silently introduce dependency to stdio. On
>>> MinGW-w64 for example, any snprintf function call will bloat binary size
>>> for 70kb, that is even we shared linking with both msvcrt and libstdc++.
>>>
>>> I frequently see a random bloat because of that.
>>>
>>> By the way, it also hurts optimizations since compilers cannot merge them
>>> together.
>>>
>>> Better just throw an exception with “vector<T> out of bounds” instead of
>>> formatting subscriptor.
>>>
>>
>>
>>I can make that depend on the _GLIBCXX_VERBOSE macro.
>
>Where is the dependency on stdio coming from though? Formatting the
>exception string doesn't use stdio.
>



  reply	other threads:[~2021-05-20 19:47 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-15  2:25 unlvsur unlvsur
2021-05-15  8:23 ` Jonathan Wakely
2021-05-20 19:20   ` Jonathan Wakely
     [not found]     ` <DM6PR05MB4697628C7E6769B8FCDA1D6ED62A9@DM6PR05MB4697.namprd05.prod.outlook.com>
2021-05-20 19:35       ` Jonathan Wakely
2021-05-20 19:47         ` unlvsur unlvsur [this message]
2021-05-20 19:50           ` unlvsur unlvsur
2021-05-20 20:00             ` Jonathan Wakely
2021-05-20 19:52           ` Jonathan Wakely

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=DM6PR05MB469776570EE332423E1BE7F9D62A9@DM6PR05MB4697.namprd05.prod.outlook.com \
    --to=unlvsur@live.com \
    --cc=jwakely@redhat.com \
    --cc=libstdc++@gcc.gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).