[fiber, afio, dl, bindlib, expected, hana, tick, fit, simd, dispatch] Checking my shortlist of upcoming Boost C++ 11/14 librari
Dear Boost, I've spent the last few days reviewing the shortlist of C++ 11/14 libraries as suggested in this thread http://article.gmane.org/gmane.comp.lib.boost.devel/255548 which appear closest to entering Boost as per the rules in my C++ Now abstract: 1. The library must be C++ 11/14 only. As in, no C++ 03 compilation possible. 2. The order chosen shall be the closest to entering Boost in March/April (i.e. now), so already in Boost, approved to enter Boost, in the formal review queue, approaching the formal review queue. At https://drive.google.com/open?id=1Bizs9xz2dyt1d5GJv-pOCmu1cS67gRd0KNYn sjYK Ji w&authuser=0 you will hopefully find a traffic light dashboard spreadsheet (I tried posting a PNG here, it was refused). In it is shown the current state, as estimated by fallible me from a cursory glance at the library, for various measures of the library. Errors are entirely possible, hence me asking here before I make a fool of myself. List members may be as surprised as I was to learn that most of the upcoming libraries require C++ 14. They are also universally standalone from Boost, being about as weakly dependant as is possible and still be called a Boost library. That was a big surprise to me as well. Another surprise was the dislike of Boost.Test - a lot of libraries don't use it, indeed can't use it. And finally, almost everyone uses Travis CI, though not valgrind memcheck for leak detection nor coveralls for unit test coverage. The first category is libraries already in the formal review queue or better. These are: Fiber, AFIO, Dl. The second category is libraries where the authors have been visible enough for me to review the library and score it onto the dashboard. These are: BindLib, Expected, Hana, Tick, Fit. ... of which BindLib, Expected and Hana are the most "Boost ready" by my estimation. I still haven't reviewed Sqlpp11, SIMD, Dispatch, Range v3, Spirit X3 nor Proto 0x. The main reason why not is that I would never use any of those libraries in any work I expect to do anytime in the next five years. That's quite unfair to those libraries, but I have another big reason: I'm also hesitant to review libraries covered by their authors also present at C++ Now, or have presented on them in the past, as it won't add value plus I probably wouldn't know what I'm talking about compared to them so that removes Expected?, Hana, SIMD, Dispatch, Range v3, Spirit X3 and Proto 0x. Which leaves a shortlist of: Fiber, AFIO, Dl, BindLib, maybe Expected, maybe Tick and Fit. I think SIMD and Dispatch have been presented on before at C++ Now right? Could anyone also very quickly summarise the main differences in their C++ 11 only rewrite? I think I remember it was mainly compilation speed rather than doing anything unusually C++ 11 only? Niall -- ned Productions Limited Consulting http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/
On 3 Apr 2015 at 1:57, Niall Douglas wrote:
At https://drive.google.com/open?id=1Bizs9xz2dyt1d5GJv-pOCmu1cS67gRd0KNYn sjYK Ji w&authuser=0 you will hopefully find a traffic light dashboard spreadsheet (I tried posting a PNG here, it was refused).
Stupid URL has added spaces it in :( Try this shortened one instead: http://goo.gl/yeE9qX Niall -- ned Productions Limited Consulting http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/
On Fri, Apr 3, 2015 at 2:57 AM, Niall Douglas
Dear Boost,
I've spent the last few days reviewing the shortlist of C++ 11/14 libraries as suggested in this thread http://article.gmane.org/gmane.comp.lib.boost.devel/255548 which appear closest to entering Boost as per the rules in my C++ Now abstract:
1. The library must be C++ 11/14 only. As in, no C++ 03 compilation possible.
2. The order chosen shall be the closest to entering Boost in March/April (i.e. now), so already in Boost, approved to enter Boost, in the formal review queue, approaching the formal review queue.
At https://drive.google.com/open?id=1Bizs9xz2dyt1d5GJv-pOCmu1cS67gRd0KNYn sjYK Ji w&authuser=0 you will hopefully find a traffic light dashboard spreadsheet (I tried posting a PNG here, it was refused). In it is shown the current state, as estimated by fallible me from a cursory glance at the library, for various measures of the library. Errors are entirely possible, hence me asking here before I make a fool of myself.
Fixed link: https://docs.google.com/spreadsheets/d/1Bizs9xz2dyt1d5GJv-pOCmu1cS67gRd0KNYn...
List members may be as surprised as I was to learn that most of the upcoming libraries require C++ 14. They are also universally standalone from Boost, being about as weakly dependant as is possible and still be called a Boost library. That was a big surprise to me as well. Another surprise was the dislike of Boost.Test - a lot of libraries don't use it, indeed can't use it. And finally, almost everyone uses Travis CI, though not valgrind memcheck for leak detection nor coveralls for unit test coverage.
The first category is libraries already in the formal review queue or better. These are:
Fiber, AFIO, Dl.
The second category is libraries where the authors have been visible enough for me to review the library and score it onto the dashboard. These are:
BindLib, Expected, Hana, Tick, Fit.
... of which BindLib, Expected and Hana are the most "Boost ready" by my estimation.
I still haven't reviewed Sqlpp11, SIMD, Dispatch, Range v3, Spirit X3 nor Proto 0x. The main reason why not is that I would never use any of those libraries in any work I expect to do anytime in the next five years.
That's quite unfair to those libraries, but I have another big reason: I'm also hesitant to review libraries covered by their authors also present at C++ Now, or have presented on them in the past, as it won't add value plus I probably wouldn't know what I'm talking about compared to them so that removes Expected?, Hana, SIMD, Dispatch, Range v3, Spirit X3 and Proto 0x. Which leaves a shortlist of:
Fiber, AFIO, Dl, BindLib, maybe Expected, maybe Tick and Fit.
I think SIMD and Dispatch have been presented on before at C++ Now right? Could anyone also very quickly summarise the main differences in their C++ 11 only rewrite? I think I remember it was mainly compilation speed rather than doing anything unusually C++ 11 only?
Niall
-- ned Productions Limited Consulting http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
On 03/04/2015 02:57, Niall Douglas wrote:
I think SIMD and Dispatch have been presented on before at C++ Now right? Could anyone also very quickly summarise the main differences in their C++ 11 only rewrite? I think I remember it was mainly compilation speed rather than doing anything unusually C++ 11 only?
Boost.Dispatch underwent a c++11 rewrite for compile-time issue yes, the main point being to use proper modern type deduction facility and be less constraining on how to provide dispatching point. Mathias Gaunard can probably be more talkative on the details. Boost.SIMD is still being updated but it's mostly cosmetic/usability changes for now. Nothing in Boost.SIMD scream c++11-only by design, it's just adapting to new pratice.
On 03/04/15 15:21, joel falcou wrote:
On 03/04/2015 02:57, Niall Douglas wrote:
I think SIMD and Dispatch have been presented on before at C++ Now right? Could anyone also very quickly summarise the main differences in their C++ 11 only rewrite? I think I remember it was mainly compilation speed rather than doing anything unusually C++ 11 only?
Boost.Dispatch underwent a c++11 rewrite for compile-time issue yes, the main point being to use proper modern type deduction facility and be less constraining on how to provide dispatching point. Mathias Gaunard can probably be more talkative on the details.
I initially wanted to do a presentation on Boost.Dispatch, and how the new versions of C++ significantly simplify what it's doing, but big changes in my personal and professional life have prevented me from considering going to C++Now this year. Maybe some other time though.
participants (4)
-
joel falcou -
Klaim - Joël Lamotte -
Mathias Gaunard -
Niall Douglas