[PATCH] D143160: [include-mapping] Introduce a human-edit CXXSymbolMapping file
Haojian Wu via Phabricator via cfe-commits
cfe-commits at lists.llvm.org
Mon Feb 6 05:54:34 PST 2023
hokein added inline comments.
================
Comment at: clang/include/clang/Tooling/Inclusions/StdAlternativeHeaderMap.inc:3
+//
+// This is a hand-curated list for C++ symbols (e.g. provided by multiple
+// headers), to address the short comings of cppreference or automated
----------------
kadircet wrote:
> hokein wrote:
> > kadircet wrote:
> > > `This is a hand-curated list for C++ symbols` reads like we're planning to put all special C++ symbols into this file, rather than just the ones that are provided by alternative headers. that's the reason why i mentioned it as `This is a hand-curated list for symbols provided by multiple headers` specifically.
> > Restricting the file only to multiple-header symbols seems a bit narrow (and the `consume_header` symbol only has one header which doesn't fit into this bucket nicely).
> >
> > My take of this file is - we'll put all special C++ symbols that are not able to handle by the cppreference generator, multiple-header symbols are the most critical ones.
> >
> > What do you think?
> >
> >
> >
> > (and the consume_header symbol only has one header which doesn't fit into this bucket nicely).
>
> Well, i'd say they deserve their own list in that case.
>
>
> > My take of this file is - we'll put all special C++ symbols that are not able to handle by the cppreference generator, multiple-header symbols are the most critical ones.
>
> I am afraid of that list getting too long and impossible to read manually any more.
> but since these are going to be private files soon, we can always do that split once that's actually the case.
>
> We need a different name for this file in that case though. As I thought we are only putting symbols with alternative headers into this file. What about `StdSpecialSymbolMap.inc` instead and update the file comment to:
> ```
> This is a hand-curated list for C++ symbols that cannot be parsed/extracted via include-mapping tool.
> ```
>
> and not talk about `All headers for a symbol name provide the same declaration (hence these are not overloads/variants like std::remove from algorithm vs cstdio).` as we're planning to add them to here as well. it's just we aren't adding them "right now".
Yeah, this sounds good to me.
> Well, i'd say they deserve their own list in that case.
Yeah, this is an alternative (a mono file vs. muti files). I would start with a single file.
The list of these symbols is not that large now.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D143160/new/
https://reviews.llvm.org/D143160
More information about the cfe-commits
mailing list