Re: [PATCH] Fix missing range information for "%q+D" format code

[David Malcolm <dmalcolm@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 2645 bytes --]

On Fri, 2015-12-04 at 12:09 +0100, Bernd Schmidt wrote:
> On 12/03/2015 09:33 PM, David Malcolm wrote:
> > The attached patch updates the handling of %q+D, simplifying
> > the implementation, and ensuring that it retains the range
> > information of the decl, giving:
> >
> > diagnostic-ranges-1.c:6:7: warning: unused variable ‘redundant’ [-Wunused-variable]
> >     int redundant;
> >         ^~~~~~~~~
> 
> For most of it I've convinced myself that it looks OK.

Thanks.

> >   void
> > -rich_location::set_range (unsigned int idx, source_range src_range,
> > -			  bool show_caret_p, bool overwrite_loc_p)
> > +rich_location::set_range (line_maps *set, unsigned int idx,
> > +			  source_location loc, bool show_caret_p)
> >   {
> 
> Here you need to update the function comment.

Is the attached suitable?

> This is a bit of a strange function. As far as I can tell, it's called 
> from only two places, one in c-common.c (which overwrites idx 0, i.e. 
> the entire range, and therefore is uninteresting), 

idx 0 is the primary location associated with the rich_location; but I'm
not sure it's strictly correct to say "the entire range".  In theory
there could be other locations/ranges within the rich_location e.g. if
someone did something like:

  rich_location richloc (loc_A); /* primary loc is "loc_A" */

  /* ...then add a couple of other ranges of interest
     e.g. to print a bogus infix binary operator,
     or somesuch.  */
  richloc.add_range (range_B);
  richloc.add_range (range_C);

  /* richloc now has:
       idx == 0 (primary) at loc_A
       idx == 1 at range_B
       idx == 2 at range_C.  */
 
  /* emit it... */
  error_at_richloc (&richloc,
                    "something bad involving %q+D", decl);

where the %q+D leads to the idx == 0 location being overwritten by
DECL_SOURCE_LOCATION (decl) instead of loc_A.

> and one in 
> text_info::set_location. I wonder about the use of "idx" in the latter. 
> As far as I can tell, that's used by the Fortran frontend to keep track 
> of two separate ranges for their diagnostics - correct?

Yes.    These format codes that manipulate the location of the
diagnostic were one of the big surprises to me when I was writing this.
This function exists to support them.

> Is that really 
> related to the normal function of rich_location and how it keeps track 
> of multiple ranges, 

Yes.

> or would that be better expressed by keeping two 
> rich_locations in text_info?

No; keeping track of multiple locations/ranges, potentially with
multiple carets is rich_location's job.


Hope this makes sense; does it look OK with the proposed comment fix?

Dave

[-- Attachment #2: comment.patch --]
[-- Type: text/x-patch, Size: 955 bytes --]

diff --git a/libcpp/line-map.c b/libcpp/line-map.c
index 333e835..0c8f328 100644
--- a/libcpp/line-map.c
+++ b/libcpp/line-map.c
@@ -2064,14 +2064,14 @@ rich_location::add_range (location_range *src_range)
   m_ranges[m_num_ranges++] = *src_range;
 }
 
-/* Add or overwrite the range given by IDX.  It must either
-   overwrite an existing range, or add one *exactly* on the end of
+/* Add or overwrite the location given by IDX.  It must either
+   overwrite an existing location, or add one *exactly* on the end of
    the array.
 
    This is primarily for use by gcc when implementing diagnostic
    format decoders e.g. the "+" in the C/C++ frontends, for handling
    format codes like "%q+D" (which writes the source location of a
-   tree back into range 0 of the rich_location).
+   tree back into location 0 of the rich_location).
 
    If SHOW_CARET_P is true, then the range should be rendered with
    a caret at its starting location.  This