Roi Barkan

C++ enthusiast

[Noy Alroy Schnitzer @noyalroy]

את האמת. בעיקר עצוב לי.
הייתה לנו הזדמנות להקים קואליציה ליברלית רחבה שמשקפת את בחירתם של רוב תושבי ת״א.
במקום זה שוב ניצחה הנקמנות הפוליטית את קול ההגיון והשפיות.
דווקא בעת הזו ואחרי כל מה שעברנו בישראל בשנה וחצי האחרונה, ת״א הייתה אמורה להיות המגדלור.

[Roi Barkan @RoiBarkan]

@seanbax @CppSage @atomgalaxy That specific UB also “applies” to functions that return ‘objects with a promise_type’ which innocent readers might think are coroutines. I learned that here: https://t.co/uzGSlmLFAh (@Superlokkus)

[Roi Barkan @RoiBarkan]

@seanbax Is it more complicated to add a use-after-move static analysis (as part of acopiler or linter) than to add the notion of destructive move to the language?

[CppCon @CppCon]

We have just released a new CppCon 2023 Video!

More Ranges Please - by @roibarkan- CppCon 2023

https://t.co/SXlAimgWEg

#Cpp20 #Cpp23 #cppcon

[Roi Barkan @RoiBarkan]

@CppCon I really enjoyed preparing and giving this talk!
Thanks a lot for the opportunity.
Direct link to the slides: https://t.co/UA5LIQxDUc.

[The Jerusalem Post @Jerusalem_Post]

The Jerusalem Post can now confirm based on verified photos of the bodies that the reports of babies being burnt and decapitated in Hamas's assault on Kfar Aza are correct.

May their memory be a blessing.
https://t.co/LaVs1aXR50

[Visegrád 24 @visegrad24]

The 90-year-old Holocaust survivor Gina Smiatich was murdered in her own home in Kibbutz Kissufim by Hamas.

She was dragged out of a shelter, taken to her living room, forced to kneel and then shot in the back of her head.

[Roi Barkan @RoiBarkan]

@hankadusikova I really like the all() function that takes a range<task<T>> and returns a task<vector<T>>. I guess functional programmers have a name for such types of functions

[Roi Barkan @RoiBarkan]

@vector_of_bool I believe this term came out as an offspring of the term “feature engineering” - considered an “inelegant” part of other ML algorithms. Deep neural networks promised to free us from ‘feature engineering’ - but then ‘prompt engineering’ emerged instead for large models.

[Roi Barkan @RoiBarkan]

@lefticus Do OTP (one time password) devices meet the criteria?

[Roi Barkan @RoiBarkan]

@fenbf Great article. I gave a talk last week on @corecpp on the topic of ranges. Here are my slides: https://t.co/58ZiFaNO1z
Comments are welcome

[C++ on Sea @cpponsea]

C++ On Sea 2023 SESSION ANNOUNCEMENT: Coroutine Intuition by Roi Barkan

https://t.co/twg1N2ctQL

Register now at https://t.co/EGtQ3aeGjK

#coroutines #cpp20 #generators #patterns #cpponsea

[Roi Barkan @RoiBarkan]

@supahvee1234 @david_sankel @corecpp Great content, indeed. I also liked that you based a lot of it on contribution that you made to SFML, and the citations/links to your empirical results

[Roi Barkan @RoiBarkan]

@vector_of_bool I think cppcoro and asio are likely the most widely used.

[Roi Barkan @RoiBarkan]

@ericniebler @zygoloid @nagyegrimate BTW, a great talk on this topic by Nina Ranns (given on @ACCUConf) just came online: https://t.co/OlTAyo0At5

[Roi Barkan @RoiBarkan]

@ericniebler @zygoloid @nagyegrimate The standard does say (https://t.co/8lR6oYiu96) that “Standard-layout classes are useful for communicating with code written in other programming languages”. I guess other c++ compilers are further away than other programming languages…

[Roi Barkan @RoiBarkan]

@seanbax @__qualname__ @pdimov2 @david_sankel Thanks Sean. As the thread mentions - it seems that the std::simd is in the works, and I'm excited on one hand and concerned about the best ways to reach something that's actually meaningful for those that need more than auto-vectorization

[Roi Barkan @RoiBarkan]

@seanbax @__qualname__ @pdimov2 @david_sankel Cheers, thanks for the quick response and fix.
I wonder if you think the ‘chunk’ approach is artificial, or what might be a common/typical way for people to try and manually leverage simd in their code

[Roi Barkan @RoiBarkan]

@seanbax @__qualname__ @pdimov2 @david_sankel I think that a mix of implicit conversion from scalar to vector and implicit conversion of 'bool vectors' to bool might lead people to unexpected results. It seems that in Circle the latter is explicit which makes things less error-prone.
Example: https://t.co/H7dwRefZmW

[Roi Barkan @RoiBarkan]

@BarryRevzin Perhaps you can add ‘sudo apt-get update —fix-missing’ around/before line 407 of range-v3-ci.yml, similar to https://t.co/YZA0Pj0EOe