Re: [Qemu-devel] [PATCH v4 37/38] raw: Convert a warning to warn_report()

[Markus Armbruster]

Kevin Wolf <address@hidden> writes:

> Am 17.10.2018 um 10:27 hat Markus Armbruster geschrieben:
>> Convert the warning about dangerous automatic probing of raw images to
>> warn_report().  Split its text to conform to conventions spelled out
>> in warn_report()'s contract.
>
> One of the things the warn_report() documentation says is:
>
>     It's wrong to call this in a QMP monitor.  Use error_setg() there.
>
> Probably a bug in the comment (copy and paste from error_report())? The
> same sentence is also there for info_report().

We could debate whether use of warn_report() in a QMP monitor is a good
idea, but the comment is definitely wrong: error_setg() can't do
warnings.  I'll fix this.

>> Update expected output of qemu-iotest 109 accordingly.  Update
>> qemu-iotest 099's output filtering to keep filtering out the warning.
>> 
>> Cc: Kevin Wolf <address@hidden>
>> Cc: Max Reitz <address@hidden>
>> Signed-off-by: Markus Armbruster <address@hidden>
>
>> --- a/tests/qemu-iotests/109.out
>> +++ b/tests/qemu-iotests/109.out
>> @@ -5,8 +5,9 @@ QA output created by 109
>>  Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=67108864
>>  Formatting 'TEST_DIR/t.raw.src', fmt=IMGFMT size=67108864
>>  {"return": {}}
>> -WARNING: Image format was not specified for 'TEST_DIR/t.raw' and probing 
>> guessed raw.
>> -         Automatically detecting the format is dangerous for raw images, 
>> write operations on block 0 will be restricted.
>> +warning: Image format was not specified for 'TEST_DIR/t.raw' and probing 
>> guessed raw
>> +Automatically detecting the format is dangerous for raw images, write
>> +operations on block 0 will be restricted.
>>           Specify the 'raw' format explicitly to remove the restrictions.
>
> I honestly like the old formatting better, with indentation and without
> the arbitrary \n in the error_printf() line that is probably shorter
> than the warning line anyway.

We don't indent an informational message following an error or a warning
elsewhere.  I'm willing to keep the indentation anyway.  You tell me.

I dislike text wider than some 70 characters.  I'm willing to keep it
that way anyway.  You tell me.

> Also, I know that qemu-iotests ignores whitespace for diffing the
> result, but not adjusting the last line in the reference output still
> looks a bit odd.

That's an accident.

Thanks!