|
Lectem
@
lectem
Bordeaux, France
|
|
French game engine developper and c++
enthusiast.
|
|
|
2.028
Tweetovi
|
358
Pratim
|
206
Osobe koje vas prate
|
| Tweetovi |
|
Lectem
@lectem
|
1. velj |
|
For those who are interested, I spent a day with Gitlab CI, then tried Azure Pipelines.
I'm not going back, Azure Pipelines is awesome!
I only found one (very) tiny default up until now, and the rest has been a joy to work with!
Now planning to create several c++ CI docker images twitter.com/lectem/status/…
|
||
|
|
||
|
Lectem
@lectem
|
1. velj |
|
I did try with everything but it was just noise :)
|
||
|
|
||
|
Lectem
@lectem
|
1. velj |
|
I'm no expert, but I think that temporary objects lifetime ends at the end of the full-expression.
And iirc, `int a = min(1,2)` is one. So if I could write `const int [a, b] = std::minmax(rand(),rand())` then it would be valid?
I could be wrong about what a full-exp is though.
|
||
|
|
||
|
Lectem
@lectem
|
1. velj |
|
Sadly none does godbolt.org/z/F3g5zr
|
||
|
|
||
| Lectem proslijedio/la je tweet | ||
|
Freya Holmér
@FreyaHolmer
|
31. sij |
|
I've decided to do the same
GDC is very inaccessible to most game developers in the world
and
they don't pay for flights
they don't pay for hotel
they don't pay for the talk, that they then have the rights for
it's not worth it anymore, I have more reach on my youtube channel twitter.com/jesawyer/statu…
|
||
|
|
||
|
Lectem
@lectem
|
1. velj |
|
Exactly...
Which makes it really hard to use std::minmax since you can't force a copy with structure bindings because auto.
I think it's actually a case against auto, there's the same issue with min and max for example.
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Mmhh, I think most of this annex breaks predate C++11 no?
I'm not sure if differences with C should still be called breaks at this point anyway. Since C11 C++ and C went in their own directions :/
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Mmhh, what code broke?
I would say they tend to be over conservative on a lot of things, and only a few breaks have been made in the previous standards
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Structure binding and C++17 immediatly feels less interesting to me.
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Guess the issue.
C++ is so damn broken sometimes I wanna puke pic.twitter.com/2OBPmDW08P
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
I know systems that do not have path separators, yet std::path does 🤷♂️
We can go and list a lot of things like this tbh. Named threads bother me quite a bit, but not as much as stack size and priority tbh.
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Just need to make it possible to be a no-OP and we're good tbh.
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Hell, even threads are not actually available everywhere
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Then why did we standardize clocks? Filesystem? IO?
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Yeah that's the usual argument made against having those, I've been reading about it for years and still don't really like it. Since native_handle is useless one always end up reimplementing std::thread if they need one of those features
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
While the implenters actually know what they use, and are allowed to change it (if they do chose so)
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
They dont work this way because most platforms implement thread names at the OS level. But the stl can wrap any kind of class or API of the OS. Think for example mingw that wraps pthread but in reality still uses win32
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Oh, do people really use std::thread for GPUs? That sounds so un practical and inefficient
|
||
|
|
||
|
Lectem
@lectem
|
31. sij |
|
Thing is, you can't really do it unless you reimplement the whole std::(j) thread. It's been like this since c++11 and native_handle doesn't help anywhere and never will unless implementers give guarantees about what type it aliases.
|
||
|
|
||
|
Lectem
@lectem
|
30. sij |
|
And yet another reason why ISO is such a fucking bad idea for a language standardization.
I can't even create an account on the french ISO org @AFNOR
And I'm hearing that now we have to pay to be a national body member ?
Fuck it!
#Cpp pic.twitter.com/2xpoZlALuf
|
||
|
|
||