[cfe-commits] [PATCH] Correct first-line indentation in preprocessor output

[Hal Finkel]

----- Original Message -----
> From: "Jordan Rose" <jordan_rose at apple.com>
> To: "Hal Finkel" <hfinkel at anl.gov>
> Cc: "Eli Friedman" <eli.friedman at gmail.com>, "llvm cfe" <cfe-commits at cs.uiuc.edu>
> Sent: Thursday, January 24, 2013 11:55:20 AM
> Subject: Re: [cfe-commits] [PATCH] Correct first-line indentation in preprocessor output
> 
> Hm. If this is the change we go with, there should definitiely be a
> comment saying that MoveToLine(SourceLocation) returns a different
> value than MoveToLine(unsigned). That, or they should be changed to
> be consistent.

Good point.

> 
> (My first inclination was to change
> PrintPPOutputPPCallbacks::HandleFirstTokOnLine, and keep the current
> semantics of MoveToLine, but then you'd have to call getPresumedLoc
> again. Or not use the convenience wrapper for MoveToLine. But it's
> probably okay since, as Eli pointed out, this is the only consumer
> of the return value.)

Exactly.

> 
> Also, you should probably add a ^ to your {{       }} in your test
> case, just to make things stricter/easier for FileCheck.

Done.

Revised patch attached. How's this?

Thanks again,
Hal

> 
> Jordan
> 
> 
> On Jan 9, 2013, at 15:00 , Hal Finkel <hfinkel at anl.gov> wrote:
> 
> > ----- Original Message -----
> >> From: "Eli Friedman" <eli.friedman at gmail.com>
> >> To: "Hal Finkel" <hfinkel at anl.gov>
> >> Cc: "llvm cfe" <cfe-commits at cs.uiuc.edu>
> >> Sent: Wednesday, January 9, 2013 4:40:18 PM
> >> Subject: Re: [cfe-commits] [PATCH] Correct first-line indentation
> >> in preprocessor output
> >> 
> >> On Wed, Jan 9, 2013 at 2:20 PM, Hal Finkel <hfinkel at anl.gov>
> >> wrote:
> >>> Currently, the -E output from clang does not produce the correct
> >>> indentation on the first line. This is because MoveToLine returns
> >>> false, and when this happens, the regular process for producing
> >>> initial indentation is skipped. This patch makes sure this does
> >>> not happen on the first line -- it is not clear to me whether
> >>> there are other circumstances where the current logic could be
> >>> problematic.
> >>> 
> >>> It looks like calling SourceManager::getPresumedLoc is a
> >>> relatively
> >>> expensive operation, so however this is fixed, I assume that we
> >>> want to minimize calls to that function.
> >>> 
> >>> Please review.
> >> 
> >> --- a/lib/Frontend/PrintPreprocessedOutput.cpp
> >> +++ b/lib/Frontend/PrintPreprocessedOutput.cpp
> >> @@ -139,10 +139,13 @@ public:
> >>                                 diag::Mapping Map, StringRef Str);
> >> 
> >>   bool HandleFirstTokOnLine(Token &Tok);
> >> -  bool MoveToLine(SourceLocation Loc) {
> >> +  bool MoveToLine(SourceLocation Loc, bool *FirstLine = 0) {
> >>     PresumedLoc PLoc = SM.getPresumedLoc(Loc);
> >> -    if (PLoc.isInvalid())
> >> +    if (PLoc.isInvalid()) {
> >> +      if (FirstLine) *FirstLine = false;
> >>       return false;
> >> +    }
> >> +    if (FirstLine) *FirstLine = PLoc.getLine() == 1;
> >>     return MoveToLine(PLoc.getLine());
> >>   }
> >>   bool MoveToLine(unsigned LineNo);
> >> 
> >> There is precisely one user of the return value of MoveToLine: the
> >> one
> >> you're modifying.  Why do we need an extra out parameter?
> > 
> > Good point, thanks! Revised patch attached -- which now changes one
> > source line ;)
> > 
> > -Hal
> > 
> >> 
> >> -Eli
> >> 
> > 
> > --
> > Hal Finkel
> > Postdoctoral Appointee
> > Leadership Computing Facility
> > Argonne National Laboratory
> > <first-line-indent-v2.patch>_______________________________________________
> > cfe-commits mailing list
> > cfe-commits at cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
> 
> 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: first-line-indent-v3.patch
Type: text/x-patch
Size: 1187 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20130124/87fb10cd/attachment.bin>