[PATCH] Lex and ignore Microsoft's #pragma warning(...)

[Reid Kleckner]

On Wed, Sep 11, 2013 at 4:12 PM, Aaron Ballman <aaron at aaronballman.com>wrote:

> > +void PrintPPOutputPPCallbacks::PragmaWarning(SourceLocation Loc,
> > +                                             StringRef WarningSpec,
> > +                                             ArrayRef<int> Ids) {
> > +  startNewLineIfNeeded();
> > +  MoveToLine(Loc);
> > +  OS << "#pragma warning(" << WarningSpec << ':';
> > +  for (int I = 0, E = Ids.size(); I != E; ++I)
> > +    OS << ' ' << Ids[I];
>
> Use an iterator instead of indexes?
>

Meh.


> > +  OS << ')';
> > +  setEmittedDirectiveOnThisLine();
>
> This doesn't seem to support the syntax where there are multiple
> specifiers.  Eg)
>
> #pragma warning(disable:1; once:2)
>

That's OK, it should produce semantically equivalent output, down to the
#line markers.

> Index: lib/Lex/Pragma.cpp
> > ===================================================================
> > +  SmallString<64> IntegerBuffer;
>
> Seems a bit huge for the numbers we'd be lexing.
>

Yeah, copy pasto.


> In the error case, do we want to diagnose what's wrong?
>

No, there are different diagnostics.


> In all of the early returns, I think you'll want to skip until you get
> to the end of the pragma or else you'll get a cascade of errors.
>

With testing this doesn't appear necessary.  Some higher level code must be
handling this.  The other pragmas also don't do this.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20130913/42be5067/attachment.html>