April 2023 - Boost - lists.preview.boost.org

Re: [boost] Boost.Compat
by Peter Dimov 30 Apr '23

30 Apr '23

Andrey Semashev wrote: > On 4/30/23 02:18, Peter Dimov via Boost wrote: > > Andrey Semashev wrote: > >> And lastly. I realize that the proposed Boost.Compat is supposed to > >> contain direct equivalents of the standard components. But what if we > >> want to improve on that? Would this be not allowed simply because > >> this would deviate from the standard? > > > > No, it wouldn't be allowed. Improving upon the standard makes it > > impossible to later replace the compatibility header with a using > > declaration. We have such a problem today with e.g. Boost.Ratio and > > Boost.Chrono, where because of extensions we can't simply make them be > <ratio> and <chrono>. > > > > It's fine to want to improve upon the standard, but Compat will not be > > the place for that. That's what dedicated libraries are for. > > This would mean more code duplication, wouldn't it? It's a tradeoff as any other. You need <latch>. Since it's trivial to implement over C++11, you add a private implementation to your library. This duplicates the one in Boost.Thread. But you don't want to depend on Boost.Thread, because if you don't need C++03 support or thread interruption, it duplicates the standard library, and adds 50 dependencies. https://pdimov.github.io/boostdep-report/boost-1.82.0/module-weights.html#w… In addition, the code duplication in Compat will gradually disappear as its minimum C++ standard level is raised. E.g. shared_mutex.hpp (assuming we have it) will effectively disappear at 14. latch.hpp will disappear at 20. We can't clean up Boost.Thread in this manner even if we impose a C++11 requirement, since it has to provide its own primitives for everything because of thread interruption. But you know this because that's what the Sync intro says. :-)

1 0

Re: [boost] Boost.Compat
by Andrey Semashev 30 Apr '23

30 Apr '23

On 4/30/23 14:58, Peter Dimov via Boost wrote: > Andrey Semashev wrote: >> I didn't say anything against refactoring, in fact I'm arguing that we should >> maintain our existing libraries (which includes refactoring) rather than add >> new libraries with duplicate components. > > You did create the (stillborn) Sync though, which duplicates Thread. Why? It doesn't fully duplicate Boost.Thread. It added new synchronization primitives and added support for different timing libraries besides Boost.Chrono. There was another idea that Boost.Thread became too large and difficult to maintain. So the thread synchronization primitives were supposed to be extracted to a separate library, Boost.Sync, leaving only thread launching and high-level tools like future/promise, async and thread pool in Boost.Thread.

1 0

Re: [boost] Boost.Compat
by Peter Dimov 30 Apr '23

30 Apr '23

Andrey Semashev wrote: > I didn't say anything against refactoring, in fact I'm arguing that we should > maintain our existing libraries (which includes refactoring) rather than add > new libraries with duplicate components. You did create the (stillborn) Sync though, which duplicates Thread. Why?

1 1

Re: [boost] Boost.Compat
by Andrey Semashev 30 Apr '23

30 Apr '23

On 4/30/23 02:18, Peter Dimov via Boost wrote: > Andrey Semashev wrote: >> And lastly. I realize that the proposed Boost.Compat is supposed to contain >> direct equivalents of the standard components. But what if we want to >> improve on that? Would this be not allowed simply because this would deviate >> from the standard? > > No, it wouldn't be allowed. Improving upon the standard makes it impossible > to later replace the compatibility header with a using declaration. We have > such a problem today with e.g. Boost.Ratio and Boost.Chrono, where because > of extensions we can't simply make them be <ratio> and <chrono>. > > It's fine to want to improve upon the standard, but Compat will not be the > place for that. That's what dedicated libraries are for. This would mean more code duplication, wouldn't it?

1 0

Re: [boost] Boost.Compat
by Vinnie Falco 30 Apr '23

30 Apr '23

On Sat, Apr 29, 2023 at 4:06 PM Andrey Semashev via Boost <boost(a)lists.boost.org> wrote: > If we don't want to use the libraries we write and accept during > the review then there must be something wrong with Boost. Nonsense. Large code bases need refactoring at regular intervals, this is no different. Or do you think that Boost libraries are so perfect upon acceptance that they can never be improved? My libraries are certainly not. Thanks

2 1

Re: [boost] Boost.Compat
by Andrzej Krzemienski 30 Apr '23

30 Apr '23

niedz., 30 kwi 2023 o 02:14 Vinnie Falco via Boost <boost(a)lists.boost.org> napisał(a): > On Sat, Apr 29, 2023 at 4:06 PM Andrey Semashev via Boost > <boost(a)lists.boost.org> wrote: > > If we don't want to use the libraries we write and accept during > > the review then there must be something wrong with Boost. > > Nonsense. Large code bases need refactoring at regular intervals, this > is no different. Or do you think that Boost libraries are so perfect > upon acceptance that they can never be improved? My libraries are > certainly not. > I think what Andrey was referring to is the situation where there exists one implementation, say in Boost.Core, and then a duplicate implementation in another Boost library for the purpose of reducing dependencies. Regards, &rzej; > > Thanks > > _______________________________________________ > Unsubscribe & other changes: > http://lists.boost.org/mailman/listinfo.cgi/boost >

1 0

Re: [boost] Boost.Compat
by Glen Fernandes 29 Apr '23

29 Apr '23

On Sat, Apr 29, 2023 at 7:24 PM Peter Dimov wrote: > Andrey Semashev wrote: > > And lastly. I realize that the proposed Boost.Compat is supposed to > contain > > direct equivalents of the standard components. But what if we want to > > improve on that? Would this be not allowed simply because this would > deviate > > from the standard? > > No, it wouldn't be allowed. Improving upon the standard makes it impossible > to later replace the compatibility header with a using declaration. We have > such a problem today with e.g. Boost.Ratio and Boost.Chrono, where because > of extensions we can't simply make them be <ratio> and <chrono>. > > It's fine to want to improve upon the standard, but Compat will not be the > place for that. That's what dedicated libraries are for. > I agree. If Boost.Compat exists then its components should not deviate from their standard counterparts. It's not the place to innovate, only to provide an implementation (if possible) for a standard library facility.

1 0

Re: [boost] Boost.Compat
by Peter Dimov 29 Apr '23

29 Apr '23

Andrey Semashev wrote: > And lastly. I realize that the proposed Boost.Compat is supposed to contain > direct equivalents of the standard components. But what if we want to > improve on that? Would this be not allowed simply because this would deviate > from the standard? No, it wouldn't be allowed. Improving upon the standard makes it impossible to later replace the compatibility header with a using declaration. We have such a problem today with e.g. Boost.Ratio and Boost.Chrono, where because of extensions we can't simply make them be <ratio> and <chrono>. It's fine to want to improve upon the standard, but Compat will not be the place for that. That's what dedicated libraries are for.

1 0

Re: [boost] CI Test time
by Andrey Semashev 29 Apr '23

29 Apr '23

On 4/29/23 23:45, Robert Ramey via Boost wrote: > Looks like the CI for the serialization library takes about 200 minutes > to run (on a good day). When I run it on my local desktop for one > compiler/configuration it takes 1-2 minutes. It seems to me that the > way that CI works, it entails a huge waste resources and time. Has > anyone thought about this? If you know where most overhead is coming from, you can work towards reducing that overhead. For example, you can save time wasted on environment setup and git checkout if you run tests for different configurations in one CI job. If you want more specific comments then you probably have to provide more details on your problem.

1 0

Re: [boost] Boost.Compat
by Andrey Semashev 29 Apr '23

29 Apr '23

On 4/27/23 18:16, Peter Dimov via Boost wrote: > Newer C++ standards (20) have been adding useful things to the > standard library (<bit>, <latch>, etc) and most of us would like to > use them, both in Boost and outside of it, while still supporting > C++11, or 14, or 17. > > Many of these components are straightforward to implement > using the C++11 standard library, consist of not much than one > relatively small header, don't need many Boost dependencies > beyond the required ones (Assert, ThrowException, Config), and > since their interface is already fixed by the standard, don't need > a design review. > > It would be nice if we had a dedicated place in Boost where to > put these, to avoid the need for everyone reimplementing a > private copy. > > At the moment, some of these have gone into Core, for lack of > anywhere else, and we now have a <latch> contribution (which I > encouraged): > > https://github.com/boostorg/core/pull/144 > > However, Core is not quite the right place for these components. > > So my suggestion is to create a separate library/repo, Boost.Compat > (boostorg/compat, namespace boost::compat) for them. > > Thoughts? (Be sure to read the discussion in the PR as well.) My main concern in the aforementioned PR was increased code duplication, and it isn't fixed by having Boost.Compat. I really don't like us having multiple copies of the same components for the sole purpose of reducing dependencies (in the case of latch - reducing tests dependencies, not even library dependencies). If we don't want to use the libraries we write and accept during the review then there must be something wrong with Boost. This isn't good for Boost (more maintenance required, less overall quality because some parts are left unmaintained) and this isn't good for users (more confusion, more compatibility issues). Secondly, I'm not sure I like the idea of another kitchen sink kind of library a-la Boost.Core and Boost.Utility, as such libraries tend to become bloated over time. Such libraries are also inconvenient for users' consumption, as it is more difficult to find a component in one of such libraries compared to in one of the specialized ones. I would prefer us to keep evolving our existing libraries, where appropriate, and accept new ones, where their functionality is not covered by the existing ones. And lastly. I realize that the proposed Boost.Compat is supposed to contain direct equivalents of the standard components. But what if we want to improve on that? Would this be not allowed simply because this would deviate from the standard? If allowed, would this still not require a review? I think, the "no review required" policy might be a bit too permissive.

1 0