Re:C++ Workflow: C++ Parallel Computing and Asynchronous Net...
by Barenboim on Monday May 24, 2021 @01:35AM
(#27261371)
At first we used pthread only. But after the windows branch was added, we have to make it compatible with windows and standard <mutex> was used in the some shared codes.
Codes in the "kernel" folder are for Linux only, and some of them are C.
Re:C++ Workflow: C++ Parallel Computing and Asynchronous Net...
by jcelerier on Sunday May 23, 2021 @04:51PM
(#27258479)
How is that an advantage ? either you have N overloads doing manually what templates do automatically, or you have indirection and the implied performance loss
Re:C++ Workflow: C++ Parallel Computing and Asynchronous Net...
by 10000truths on Sunday May 23, 2021 @06:52PM
(#27259368)
Those few seconds can add up very quickly when you’re making incremental changes and have to keep recompiling. I ended up having to write non-template wrappers for all my ASIO stuff and compile them as a separate object file, in order to keep my workflow manageable. Building that one object file would double my total build time.
Re:C++ Workflow: C++ Parallel Computing and Asynchronous Net...
by jcelerier on Sunday May 23, 2021 @07:03PM
(#27259433)
I've actually done that all day long, at most it takes 3-4 seconds between code editing and the app / test running, I'd definitely not trade faster build times for something else in that case, especially when my IDE highlights errors when I type which greatly reduces the need for constant rebuilds.
Re:C++ Workflow: C++ Parallel Computing and Asynchronous Net...
by Barenboim on Monday May 24, 2021 @04:25AM
(#27262141)
Yes. We have applications that use all this clients. This framework supports the backend development of a pretty big industrial search engine, and we tend to build very large backend applications. And as a very light-weighted library, it's also used as a simple asynchronous MySQL client, or a kafka client. And the most often, an http server or client.
Tbh, I don't think it is comparable to this project. A better alternative for Java would be Quarkus or Micronaut, ones that can compile to native using GraalVM.
Re:C++ Workflow: C++ Parallel Computing and Asynchronous Net...
by winrid on Monday May 24, 2021 @04:33PM
(#27269369)
The reason I did this comparison was to try to better understand what is in this project as I may consider using it, not to compare to Java performance wise.
For example src/manager/DNSCache.h use std::mutex but src/kernel/Executor.h use pthread_mutex_t?
That's an interesting selection. Do you have particular problem spaces in mind?
https://sparkjava.com/
Tbh, I don't think it is comparable to this project. A better alternative for Java would be Quarkus or Micronaut, ones that can compile to native using GraalVM.
I really do need to play with GraalVM.