[Boost] [Review] [Boost.MySQL] - Review Complete
Dear All, The submission of Boost.MySQL has created plenty of "lively debate". Unsurprisingly, a library which is based on Asio, of which there are few experts, has given rise to a great deal of questioning around performance, memory re-use, correct return types and so on. It has been noted that Reuben, the library's author, has been very responsive. Both in this mailing list and in terms of raising PRs. A specific concern was raised about the name of the library, and whether it infringed any commercial trade marks. Reuben took the trouble to reach out to Oracle to check. They have responded that there is no infringement and we may use the name. In the end though, despite a great deal of words written raising concerns, ideas and suggestions, we received size reviews, all of which were high quality. The results were: Accept: 3 Conditional Accept: 2 Reject: 1 Having carefully weighed up: - The stated aims of the library - The quality of the implementation - The demonstrated responsiveness of the author - The value of the library to users of MySQL - The value of the library as a teaching tool for implementing a complex, cancellable, asynchronous protocol on ASIO, which correctly adapts to the use of user-selection completion tokens, It seems right to ACCEPT this library into Boost. My congratulations to Reuben. -- Richard Hodges hodges.r@gmail.com office: +44 2032 898 513 home: +376 861 195 mobile: +376 380 212
On Wednesday, June 15, 2022, Richard Hodges wrote:
It seems right to ACCEPT this library into Boost.
My congratulations to Reuben.
Congratulations Reuben, and thank you Richard for managing the review. Since the acceptance above is unconditional, were the conditions in the two Conditional Accept reviews already satisfied, or deemed unnecessary? Glen
On Wed, 15 Jun 2022 at 21:03, Glen Fernandes via Boost < boost@lists.boost.org> wrote:
On Wednesday, June 15, 2022, Richard Hodges wrote:
It seems right to ACCEPT this library into Boost.
My congratulations to Reuben.
Congratulations Reuben, and thank you Richard for managing the review.
Since the acceptance above is unconditional, were the conditions in the two Conditional Accept reviews already satisfied, or deemed unnecessary?
I felt that Reuben sufficiently engaged with the conditions to either explain why they were not blockers or resolve them.
Glen
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
-- Richard Hodges hodges.r@gmail.com office: +44 2032 898 513 home: +376 861 195 mobile: +376 380 212
Hi all, First of all, thanks Richard for managing the review and thanks to all who took the time to write a review.
Since the acceptance above is unconditional, were the conditions in the two Conditional Accept reviews already satisfied, or deemed unnecessary?
I've taken the following as acceptance conditions: - Returning resultsets by reference to allow memory reuse. To make things easier, resultset and prepared_statement's will be converted to plain objects and their I/O operations will be moved to connection. Addressed by https://github.com/anarthal/mysql/issues/59 https://github.com/anarthal/mysql/issues/73 - Rework the read_many and read_all functions to allow for memory reuse. This will imply the introduction of a row_collection object and its non-owning counterpart row_collection_view, replacing the vectors. read_one will stay similar, read_many as it is now disappears, a new read_some will be introduced, and read_all will stay similar but allowing memory reuse. An improvement in the internal buffering strategy will be made as part of this task, too. Tracked by https://github.com/anarthal/mysql/issues/58 https://github.com/anarthal/mysql/issues/32 - Rework the value class to address all the issues raised in the review. It will stay non-owning and will be renamed to something that makes clear that it's a view. The underlying variant won't be exposed anymore, get_optional and get_std_optional will be removed in favor of something like get_if, conversions and operators < and > will likely go away, and a level of special handling for nullable values will be introduced in get. Addressed by https://github.com/anarthal/mysql/issues/85 https://github.com/anarthal/mysql/issues/80 - Support COM_PING. Tracked by https://github.com/anarthal/mysql/issues/62 Since some of these issues are not trivial at all, it may take a couple releases before I've got this ready. Regards, Ruben.
On Wed, 15 Jun 2022 at 20:42, Richard Hodges via Boost
Dear All,
The submission of Boost.MySQL has created plenty of "lively debate". [...]
It seems right to ACCEPT this library into Boost.
My congratulations to Reuben.
Ruben, congratulations! Richard, thank you very much! Best regards, -- Mateusz Loskot, http://mateusz.loskot.net
Congratulations Reuben! Thank you, Richard! I'm looking forward to the library. Em qua., 15 de jun. de 2022 às 15:42, Richard Hodges via Boost < boost@lists.boost.org> escreveu:
Dear All,
The submission of Boost.MySQL has created plenty of "lively debate".
Unsurprisingly, a library which is based on Asio, of which there are few experts, has given rise to a great deal of questioning around performance, memory re-use, correct return types and so on.
It has been noted that Reuben, the library's author, has been very responsive. Both in this mailing list and in terms of raising PRs.
A specific concern was raised about the name of the library, and whether it infringed any commercial trade marks. Reuben took the trouble to reach out to Oracle to check. They have responded that there is no infringement and we may use the name.
In the end though, despite a great deal of words written raising concerns, ideas and suggestions, we received size reviews, all of which were high quality.
The results were: Accept: 3 Conditional Accept: 2 Reject: 1
Having carefully weighed up: - The stated aims of the library - The quality of the implementation - The demonstrated responsiveness of the author - The value of the library to users of MySQL - The value of the library as a teaching tool for implementing a complex, cancellable, asynchronous protocol on ASIO, which correctly adapts to the use of user-selection completion tokens,
It seems right to ACCEPT this library into Boost.
My congratulations to Reuben.
-- Richard Hodges hodges.r@gmail.com office: +44 2032 898 513 home: +376 861 195 mobile: +376 380 212
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
-- Alan Freitas https://github.com/alandefreitas
participants (5)
-
Alan de Freitas -
Glen Fernandes -
Mateusz Loskot -
Richard Hodges -
Ruben Perez