[PATCH] D29868: Recover more gracefully when __declspec is not supported as a keyword

[David Majnemer via Phabricator via cfe-commits]

majnemer added inline comments.


================
Comment at: lib/Parse/ParseDecl.cpp:2989
+
+          Diag(Loc, diag::err_ms_attributes_not_enabled);
+          continue;
----------------
aaron.ballman wrote:
> majnemer wrote:
> > aaron.ballman wrote:
> > > compnerd wrote:
> > > > aaron.ballman wrote:
> > > > > compnerd wrote:
> > > > > > I think that we want to emit the diagnostic even if there is no parenthesis as `__declspec` is a reserved identifier, and we would normally diagnose the bad `__declspec` (expected '(' after '__declspec').
> > > > > Yes, but it could also lead to a rejecting code that we used to accept and properly handle when __declspec is an identifier rather than a keyword. e.g.,
> > > > > ```
> > > > > struct __declspec {};
> > > > > 
> > > > > __declspec some_func(void);
> > > > > ```
> > > > > By looking for the paren, we run less risk of breaking working code, even if that code abuses the implementation namespace (after all, __declspec it not a keyword in this scenario).
> > > > But we would reject that code under `-fdeclspec` anyways.  I think having the code be more portable is a bit nicer.
> > > After discussing in IRC, I decided that I agree with @compnerd on this and have changed the patch accordingly.
> > What if somebody wants to use __declspec and are using the compiler in a freestanding mode? Also, we aren't the only member of the implementor's namespace.
> Users using __declspec in a freestanding mode is a concern that I share. However, I imagine there are *far* more users who accidentally forget to pass `-fdeclspec` than there are users who are writing code in freestanding mode that wish to use `__declspec` as an identifier in a situation that proves problematic. That being said, do you think the approach in the patch would work with a warning rather than an error? I went with an error because I felt that a warning would require tentative parsing to be properly implemented, which felt like a heavy solution for a problem I didn't think anyone would run into in practice.
> 
> I'm not overly sympathetic to others in the implementor's namespace who use `__declspec` but not to implement attributes under that name. However, I could be convinced to be sympathetic if there was some conflict in practice. Do you have a case in mind?
I suppose what you've got should be fine in practice. Heck, we used to have __declspec attributes available all the time...


https://reviews.llvm.org/D29868