Re: [PING][PATCH 2/4] remove %G and %K from calls in front end and middle end (PR 98512)

[Aldy Hernandez <aldyh@redhat.com>]

On 7/1/21 10:14 PM, Martin Sebor wrote:
> On 6/30/21 5:35 PM, David Malcolm wrote:
>> On Wed, 2021-06-30 at 13:45 -0600, Martin Sebor wrote:
>>> On 6/30/21 9:39 AM, Martin Sebor wrote:

>>> @@ -90,8 +90,8 @@ NOIPA void warn_g2 (struct A *p)
>>>     g2 (p);
>>>   }
>>> -// { dg-message "inlined from 'g2'" "" { target *-*-* } 0 }
>>> -// { dg-message "inlined from 'warn_g2'" "" { target *-*-* } 0 }
>>> +// { dg-message "inlined from 'g2'" "note on line 93" { target *-*-* 
>>> } 0 }
>>> +// { dg-message "inlined from 'warn_g2'" "note on line 94" { target 
>>> *-*-* } 0 }
>>
>> You've added descriptions to disambiguate all of the various directives
>> on line 0, which is good, but I don't like the use of line numbers in
>> the descriptions, since it will get very confusing if the numbering
>> changes.
>>
>> Would it work to use the message text as the description e.g.
>>
>>    // { dg-message "inlined from 'warn_g2'" "inlined from 'warn_g2'" { 
>> target *-*-* } 0 }
>>
>> or somesuch?
> 
> It would certainly work, they're just informational labels printed
> by DejaGnu when the assertions fail.  I added them to help me see
> what they went with while working with the test.  I'm not concerned
> about the line numbers changing.  If they do and someone notices,
> they can update them, the same way they might want to if they
> rename the functions they're inlined into.

I agree with David.  Having too specific tests adds an extra maintenance 
burden.  You may not mind updating all the line numbers when anything 
changes, but others may certainly mind.

Aldy