[libcxx-commits] [PATCH] D105082: [libcxx][NFC] removes header synopses

Tue Jun 29 11:10:22 PDT 2021

cjdb added a comment.

In D105082#2847837 <https://reviews.llvm.org/D105082#2847837>, @Mordante wrote:

> In D105082#2847492 <https://reviews.llvm.org/D105082#2847492>, @cjdb wrote:
>
>> In D105082#2847148 <https://reviews.llvm.org/D105082#2847148>, @ldionne wrote:
>>
>>> The purpose of those synopses is not to document what the Standard says, but instead to document what the state of our library implementation is. As such, they are often useful to figure out what has been implemented and what's missing from the library. They are especially useful when trying to figure out what parts of a partially implemented paper are missing (for the numerous papers where we only have "In Progress" in our tables). I also find that they help when reviewing patches, since they act as a summary of what the patch is doing (at least in most cases where we're implementing something new).
>>
>> I'm not really seeing how they document the state of the library implementation. If one wants to know if something is implemented, we can just grep for it? If this info is genuinely valuable (and I'm seeing your point about "In Progress"), then I think it would be worth us moving the synopses out of the code and into status documents. I think that would be immensely more useful than having them in the same document, because they're more accessible (we can link "In Progress" statuses to specific webpages), and because it's not duplicating the interface in the code.
>
> I'm not against moving to more detailed status pages instead of a synopsis. But then we should make sure that we add these pages for the projects in a partly done stage, instead of only removing this information.

Right, by "status page", I mean "cut the synopses out of the headers and paste them into `*.rst` files verbatim" as a first step.

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D105082/new/

https://reviews.llvm.org/D105082