Re: [PATCH 1/3] Fix "b f(std::string)", always use DMGL_VERBOSE

[Pedro Alves <pedro@palves.net>]

On 2022-05-05 22:49, Carl Love wrote:
> Pedro:
> 
> I pulled down a fresh copy of gdb and tried to apply the first patch. 
> There were issues with patching the files:
> 
>  patch -p 1 < Fix__b_f_std_string__,_always_use_DMGL_VERBOSE_1.patch
> (Stripping trailing CRs from patch; use --binary to disable.)
> patching file gdb/cp-name-parser.y
> (Stripping trailing CRs from patch; use --binary to disable.)
> patching file gdb/cp-support.c
> (Stripping trailing CRs from patch; use --binary to disable.)
> patching file gdb/cp-support.h
> patching file gdb/testsuite/gdb.cp/break-f-std-string.cc (renamed from
> gdb/testsuite/gdb.cp/no-dmgl-verbose.cc)
> (Stripping trailing CRs from patch; use --binary to disable.)
> patching file gdb/testsuite/gdb.cp/break-f-std-string.exp
> (Stripping trailing CRs from patch; use --binary to disable.)
> patching file gdb/testsuite/gdb.cp/no-dmgl-verbose.exp
> Reversed (or previously applied) patch detected!  Assume -R? [n] n
> Apply anyway? [n] n
> Skipping patch.
> 1 out of 1 hunk ignored -- saving rejects to file
> gdb/testsuite/gdb.cp/no-dmgl-verbose.exp.rej
> 
> The issue seems to be with deleting the file: 
>     gdb/testsuite/gdb.cp/no-dmgl-verbose.exp
> 
> I manually deleted it.  I then used git to commit the patch to my local
> git directory so I could add the second patch.

This looks like something on your end.  I sent the patches using git send-email,
which typically is the best way to send emails in the right format.  To check,
I downloaded the whole email from the copy I got via the list (in Thunderbird for
example, just save the email to a file), and used "git am", and it applied cleanly:

 $ git reset --hard upstream/master
 ...
 $ git am /tmp/patch1.mbox
 Applying: Fix "b f(std::string)", always use DMGL_VERBOSE

As a second experiment, I also tried copy/pasting the patch to a patch file,
and apply that with "patch" as you're doing, and it applies cleanly:

 $ git reset --hard upstream/master
 ...
 $ patch  -p1 -i /tmp/patch.diff
 patching file gdb/cp-name-parser.y
 patching file gdb/cp-support.c
 patching file gdb/cp-support.h
 patching file gdb/testsuite/gdb.cp/break-f-std-string.cc (renamed from gdb/testsuite/gdb.cp/no-dmgl-verbose.cc)
 patching file gdb/testsuite/gdb.cp/break-f-std-string.exp
 patching file gdb/testsuite/gdb.cp/no-dmgl-verbose.exp

> 
> The second patch applied cleanly.
> 
> The third patch applied cleanly.
> 
> I ran the test gdb.cp/break-f-std-string.exp.  It runs without errors
> on a Power 10 system.
> 
>   testcase gdb/testsuite/gdb.cp/break-f-std-string.exp completed in 1  
>   seconds
> 
>                 === gdb Summary ===
> 
>   # of expected passes            16
> 
> I also ran the full regression suite with and without the patch.  No
> regression failures were found.
> 
> Other than the patch not cleanly removing gdb/testsuite/gdb.cp/no-dmgl-
> verbose.exp it all looks good.
> 
> Thanks for fixing this test.

Thanks for testing.