Thread #108611025
HomeIndexCatalogAll ThreadsNew ThreadReply
H
File: cpp.png (11.2 KB)
11.2 KB
11.2 KB PNG
Gabriel Dos Reis is significantly involved with C++ profiles. Which probably means that the profiles feature will turn out poorly or worse, considering Gabriel Dos Reis' record on features like modules.

Given how manipulative and dishonest Gabriel Dos Reis is, I cannot get myself to finish this talk he had on profiles earlier this year.

https://youtube.com/watch?v=Z6Nkb1sCogI

Can anyone tell me whether C++ profiles will require runtime checks in some cases?
If yes, it would complicate code generation in compilers and make programs slower.
If no, it could limit the analysis that can be done, it could complicate it, and could make it slow at compile time.

I could see the latter option being worth it even for only partial checking, like there potentially being some sweet spots in the design space that would yield decent coverage and relatively few false positives, at a relatively low cost regarding compilation times and compiler implementation complexity. But partial checking is probably not what some are hoping for.

Will it be on a per-profile basis whether a profile can require or cause runtime checks?
+Showing all 10 replies.
>>
>>108611025
>Can anyone tell me whether C++ profiles will require runtime checks in some cases?
>If yes, it would complicate code generation in compilers and make programs slower.

obviously yes
>>
>>108611025
C++ "profiles" was a vapordesign diversion from bjarne as a part of his COPE armor from his "i'm in full denial about rust's mere existence" phase.
even if something gets thrown together eventually, it will neither actually work, nor anyone will use it (you would be lucky if anyone even bothers to half-implement the useless thing).
the "modern C++" ever meme is dead, and the committee has been in their "let's grab our last bureaucracy checks" mode for some time already. and while appearances must stay up for a while, they secretly moved on already, and so should you.
>>
i feel the need to say something racist, but i have no idea how this can even be implemented. an example he used was a dangling pointer for a string_view, but how can this be implemented without a runtime? can these features even be implemented at compile time?
>>
>Which probably means that the profiles feature will turn out poorly or worse
you could've guessed it by the fact that "C++ profiles" has "C++" in the name
>>
>>108612649
The same way Rust does it. With a borrow checker
>>
>>108611025
>https://youtube.com/watch?v=Z6Nkb1sCogI
playback speed 1.25 at least

optionally brightness +33% in the dark tones ;)
>>
>>108614331
bjarne the nocoder's "profiles" were introduced as a vapordesign (like vaporware, but at the design level) concept specifically to stifle any potential workable solution that would change the language and potentially break backwards compatibility. because bjarne understood this would bring an abrupt end to his committee grift.
so no, it wouldn't be a borrow checker, because vapor profiles are the anti-borrow-checker proposition.
it's all hopeless in any case of course. and the meme is already dead. but that was the backstory.
>>
>>108615216
>because bjarne understood this would bring an abrupt end to his committee grift.
wut?
>>
>>108615237
augmented safe c++ != c++ (breaking changes like lifetime annotation, ....)
------
^^ this will lead to two things ^^
1. an admission that non-augmented c++ is unsafe. bjarne prefers to stay in denial and do retarded things like using 10 year old logic, like "well, full safety is unattainable/impossible, so there is no point in trying guaranteed memory safety".
2. "safe c++" not having the backward compatibility/XX years old ecosystem argument anymore. at that point, the fact that rust is a superior proposition in every single aspect becomes clearer living-in-denial 10 year old bjarne doesn't know, or pretends to not know that this is something everyone with a brain already realized years ago.
------
so bjarne's idea of a diversion (conscious or subconscious, doesn't matter) was to propose "profiles" as something that should work... and that's it. no details about how this could become a workable solution or any other details or even outlines of a real design.
now years have passed, and some people are apparently still trying to unearth, untangle, and discover that ingenuity that must have sparked inside bjarne's head at that time lol
>>
>>108615989
>so bjarne's idea of a diversion (conscious or subconscious, doesn't matter) was to propose "profiles" as something that should work... and that's it. no details about how this could become a workable solution or any other details or even outlines of a real design.
>now years have passed, and some people are apparently still trying to unearth, untangle, and discover that ingenuity that must have sparked inside bjarne's head at that time lol
ok, let's do better.

some idea is here:
https://www.circle-lang.org/draft-profiles.html

Reply to Thread #108611025


Supported: JPG, PNG, GIF, WebP, WebM, MP4, MP3 (max 4MB)