Why is the btree's size_type signed?

[bobaoCrbs]

Hi,
I am just wondering why the size() of btree containers is of type "signed size_t". (instead of just size_t?).
i.e. this line

Thanks
Bo

[tituswinters]

https://google.github.io/styleguide/cppguide.html#Integer_Types At the C++ Standards level roughly 2/3 of committee members agree that unsigned-sizes was a mistake, but the C++ standard is too brittle to ever change that. Non-standard types are faced with "Follow the pattern that we believe is wrong" or "Do a higher-friction thing that is a better long-term outcome." (Neither answer is great.)

…

On Fri, Jul 30, 2021 at 5:20 PM bobaoCrbs ***@***.***> wrote: Hi, I am just wondering why the size() of btree containers is of type "signed size_t". (instead of just size_t?). i.e. this line <https://github.com/abseil/abseil-cpp/blob/5de90e2673bd0c19de67c68e2fe543444dc1114a/absl/container/internal/btree.h#L241> Thanks Bo — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#993>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AC7CRX5DXF36TQDFPIXMC53T2MJRZANCNFSM5BJHS4HA> .

-- *If you get an email from me outside of the 9-5 it is *not* because I'm always on or expect an immediate response from you; it is because of work flexibility <http://www.inc.com/john-boitnott/how-flexible-hours-can-create-a-better-work-life-balance.html> . Evening and weekend emails are a sign I allocated some regular working hours for other things (such as family, gym, friends,...). And I encourage you to feel free to do the same.

[bobaoCrbs]

Thanks for the explanation. In that case, should we at least make it consistent across the absl container library? The hash map containers are “using size_type = size_t" (e.g., this line)

[mbxx]

Changing the type of btree's size from signed to unsigned would be a substantial undertaking for Google's codebase. Every place the size is used would need to be touched to make sure that the change in return type doesn't introduce a bug.

This isn't impossible. We made a similar but larger change for the return type of StringPiece::size() when we changed absl::string_view's API to match that of std::string_view. It is a lot of work, though.

Complicating matters, we don't have a great way to help our open source users handle this change. It's possible to write a Clang refactoring tool to help with the migration, but users who are either unaware of the change or don't use the refactoring tool are liable to have a really bad time.

The inconsistency is definitely unfortunate, but fixing it would be a lot of work and involve a significant amount of risk for very little upside.

[bobaoCrbs]

Make sense. Thanks for the explanation.