Re: [PATCH 4/5] gdb/testsuite: change newline patterns used in gdb_test

[Andrew Burgess <aburgess@redhat.com>]

Tom de Vries <tdevries@suse.de> writes:

> On 5/2/23 13:13, Andrew Burgess wrote:
>> Tom de Vries <tdevries@suse.de> writes:
>> 
>>> On 5/1/23 16:33, Andrew Burgess wrote:
>>>> Tom de Vries <tdevries@suse.de> writes:
>>>>
>>>>> On 3/31/23 22:20, Andrew Burgess via Gdb-patches wrote:
>>>>>> This commit makes two changes to how we match newline characters in
>>>>>> the gdb_test proc.
>>>>>
>>>>> Hi,
>>>>>
>>>>> the -wrap used in gdb_test_multiple is defined in terms of gdb_test
>>>>> semantics, but it doesn't seem to have been updated to match the new
>>>>> behaviour in gdb_test.
>>>>>
>>>>> I've filed a PR about this regression (
>>>>> https://sourceware.org/bugzilla/show_bug.cgi?id=30403 ).
>>>>
>>>> Sorry for any problems caused.  I'm not working today, but if this has
>>>> not been addressed, I'll look at this on Tuesday.
>>>
>>> AFAIU it's a silent regression, so there are no problems in term of
>>> FAILs, it's just that more work is required.
>>>
>>> FWIW, I think the root cause for introducing this regression silently is
>>> that we try to implement the same thing in two different locations, and
>>> it's just easy for things to get out of sync.  I recently fixed
>>> something similar in commit 4fa173cfd79 ("[gdb/testsuite] Fix -wrap in
>>> presence of -prompt in gdb_test_multiple"), that's why I noticed it.
>> 
>> So I believe the patch below brings gdb_test_multiple with '-wrap' back
>> into line with gdb_test.  I also updated a couple of other places that
>> used the same (old) gdb_test pattern.
>> 
>> There were nowhere near as many regressions with this change as with
>> gdb_test.  Let me know what you think.
>> 
>
> I've applied the patch and tested it, and saw no regression.  I've also 
> reviewed the patch and LGTM.

Thanks Tom,

I've gone ahead and pushed this patch -- fixing up the newline issues.

>
> However, I've now also realized that the ^ bit is missing, which was 
> also added in this patch series.
>
> In other words, say we have:
> ...
> gdb_test "print 1" "^.$decimal = 1"
> ...
> which passes fine.
>
> But then we want to annotate the PASS message with the captured 
> $decimal, and rewrite into:
> ...
> gdb_test_multiple "print 1" "" {
>      -re -wrap "^.($decimal) = 1" {
>          set var_nr $expect_output(1,string)
>          pass "$gdb_test_name (var_nr: $var_nr)"
>      }
> }
> ...
> This FAILs because ^ at the start of the pattern is not handled the same 
> way by -wrap as by gdb_test.

I'm working on an updated patch that addresses the '^' feature for
gdb_test_multiple.  I'm still testing this locally, and it will probably
be next week now before I post this -- but I will get this done, watch
this space :)

Thanks,
Andrew