This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: Fwd: [patch, fortran] Fix USE_MAPPED_LOCATION regressions
Steve,
On Wed, May 17, 2006 at 10:19:00PM -0700, Steve Kargl wrote:
>On Wed, May 17, 2006 at 09:58:57PM -0700, Per Bothner wrote:
>> Steve Kargl wrote:
>> >What is this patch fixing? I don't know what --enable-mapped-location
>> >does.
>>
>> --enable-mapped-location is an alternative way of representing
>> line number information, using the source_location type in
>> libcpp/include/line-map.h.
>>
>> The hope is --enable-mapped-location will become the default as
>> soon as various regressions are fixed, so it is important that
>> all front-ends that are built by default need to work with
>> --enable-mapped-location.
>
>Thanks for the explanation. There are a few occurences of
>lb->linenum in error.c and a few trans-*.c files. Do these
>need updating or are these lb->linenum instances initialized
>in scanner.c? If no one else gets to these questions, I'll
>try looking over the patch this weekend.
In case you intend to touch error.c, may i ask you to look over my
comment #2 to PR fortran/21061 "-Werror ignored"
(http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21061#c2) if you find the
time. I'm well aware that it is merely a cosmetic UI issue..
I only had a few questions there, specifically:
a) about the proposed testcase
b) whether to print error: or warning: (I initially did the former)
if warnings are seen and -Werror was given.