[PATCH] D133429: [llvm] Use std::size instead of llvm::array_lengthof

Thu Sep 8 08:57:54 PDT 2022

kazu added a comment.

In D133429#3777310 <https://reviews.llvm.org/D133429#3777310>, @jloser wrote:

> In D133429#3776384 <https://reviews.llvm.org/D133429#3776384>, @dblaikie wrote:
>
>> As a follow-up, probably good to deprecate the API with the necessary recommended alternative for at least a little while, to client code migrations easier.
>>
>> @kazu - this might be something you'd be interested in helping with, or pointing @jloser to the right tools/directions.
>
> Sounds good to me.  I just opened https://reviews.llvm.org/D133502 to mark the function deprecated, but let me know if you or @kazu had something else in mind.  I've fixed downstream callers in MLIR and LLDB (https://reviews.llvm.org/D133501) with Clang being on my list to do as well.  This only accounts for the call sites in LLVM codebase though rather than external downstream.  Is there a process (e.g. wait two weeks after deprecating before removal) or something of the like?

This is a great cleanup.  Thank you!

I am not aware of any documented process.  For my recent deprecation of `Optional::hasValue`, I am waiting for 3 months or so, but some call it too long.  So, wait as long as reasonable in your terms.

In any event, I think it's a good idea to clean up internal downstream users first before deprecating `array_lengthof`.  The llvm codebase and internal downstream build cleanly even with `-Werror` (with clang being the host compiler), so we don't want to introduce a lot of deprecation warnings all of a sudden.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D133429/new/

https://reviews.llvm.org/D133429