Comments (6)
rparolin
commented on May 22, 2024
2
I'd agree with @swardle.
EASTL is indeed growing but I believe the growth has been manageable thus far with not that much duplicated code. I'd say most of the recently duplicated code I have a strategy to centralize it but I avoided doing so to prevent breaking users of our most commonly used types (eg. strings and vectors). It wasn't done for any kind of runtime or compile-time performance. I will be doing that work in the future as I bring other implementations online designed for that purpose, for example of eastl::char_traits.
The thought of preprocessing our code with an external tool (despite its popularity within the games industry) doesn't thrill me. I really wouldn't even consider a solution that involved the preprocessor. I'd rather deal with standards conforming C++ code being given to a compiler and not obscure the EASTL implementations at all unless there is really a compelling reason to do so. This is a feature that the community has told us they really appreciate about our implementation and I would hate to make contributing code/fixes to the library more difficult.
from eastl.
rparolin
commented on May 22, 2024
1
Absolutely. The standard is growing and we should keep this in the back of our minds. Thanks for starting the discussion.
from eastl.
swardle
commented on May 22, 2024
1
Thanks I think I understand what you are thinking now. More like a separate preprocessor/code generator. But when it came to debugging it the might be tricky because of the extra level of indirection. Or the code might look strange because it's generated.
Also the build system for such a change is kind of annoying. We support 2 build systems for EASTL code adding complexity here probably not as easy as it might look too.
But thanks for the comment. It is always nice to hear what other people think.
Scott
… On Feb 15, 2017, at 18:57, DragoonX6 ***@***.***> wrote:
With macros I meant not one that is handled by the c++ compiler, I meant one that is processed by an external tool, sorry if that wasn't clear.
And I think having a generator "manually" inline the code is not a bad solution, especially if it were to insert comments like what I've suggested. You're right in that it can break though, it's a trade off you have to think about, but I think it can be worth it considering the growing size of EASTL.
Offtopic: I really enjoyed your talk, lots to learn from it :)
―
You are receiving this because you commented.
Reply to this email directly, view it on GitHub, or mute the thread.
from eastl.
swardle
commented on May 22, 2024
This is true I said that. There are lots of other solutions.
We could remove the duplication with Marcos for sure but this would be harder to debug for our users.
It is possible with generated code maybe we can have the best of both worlds. But it is one more tool to break.
Other stl use lots of inline functions. That do not get inlined in debug or in some cases optimized code.
I think having simple maybe some what duplicated code is not so bad. Long as you have adequate testing. But really Rob's the guy here he owns the lib and it is his job to find the right balance here.
Scott
… On Feb 15, 2017, at 15:54, DragoonX6 ***@***.***> wrote:
This was said in the talk at cppcon 2015 from Scott Wardle, where he mentioned EASTL and why it's faster in debug mode.
After which he mentions that you basically hope for the best and that you didn't introduce that many bugs.
I'm not sure how much of this is still true today, but if it is, I was thinking about a way that could get us best of both worlds.
We could set up a preprocessor that scans for a custom set of macros that would do the inlining for us.
So I was thinking that we then would end up with something comparable to this in the generated source code:
// Pre-Inlined eastl::vector<T>::push_back
// At EASTL/vector.h:123
/* insert code here*/
// End of Pre-Inlined eastl::vector<T>::push_back
So whenever you would run into a bug at one place, you can be sure you fixed it everywhere.
The obvious downside of this is that you have to generate the source code, but I think it could be worth it in the long run looking at it from a stability standpoint.
―
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub, or mute the thread.
from eastl.
DragoonX6
commented on May 22, 2024
With macros I meant not one that is handled by the c++ compiler, I meant one that is processed by an external tool, sorry if that wasn't clear.
And I think having a generator "manually" inline the code is not a bad solution, especially if it were to insert comments like what I've suggested. You're right in that it can break though, it's a trade off you have to think about, but I think it can be worth it considering the growing size of EASTL.
Offtopic: I really enjoyed your talk, lots to learn from it :)
from eastl.
DragoonX6
commented on May 22, 2024
Alright, I can respect that. Maybe this would be something to think about again later down the road.
from eastl.
Related Issues (20)
- Incorrect behaviour of vector (and other) with switched off exceptions HOT 1
- [MSVC][std:c++latest] EASTL failed to build with msvc due to error C2338 on Windows
- Eastl is susport windows kernel? HOT 1
- Building EASTL HOT 1
- [MSVC][std:c++latest] EASTL build failed due to error C2666: 'TestLruCacheInternal::Foo::operator ==': overloaded functions have similar conversions HOT 2
- intrusive_slist is not implemented
- No signed tarball for releases
- Build with Clang15 results in warnings about deprecated builtins HOT 1
- Inquiry about the expected behavior of the comparison operators in RingBuffer HOT 1
- compilation error due to `std::remove_reference` HOT 5
- Tests fail to compile: error: use of undeclared identifier 'getpid'
- Tests fail to compile: error: use of undeclared identifier 'pthread_attr_get_np'
- Tests fail to link: ld: error: undefined symbol: EA::EAMain::InitializeSignalHandler()
- Consider using reset() for fast container teardown - Best pratices HOT 2
- emscripten support? HOT 1
- [Question] Should eastl::unordered_set.reserve not check if a change is required?
- Error of RBTreeDecrement function HOT 3
- Move semantic dosn't work with eastl::list HOT 2
- Critical silent bug: vector push back of an own element moves it and calls a copy constructor on the invalidated reference HOT 1
- Overriding make_unique and default_delete HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from eastl.