On 08/22/2018 06:59 AM, Max Reitz wrote:
> On 2018-08-21 02:10, John Snow wrote:
>>
>>
>> On 08/17/2018 03:04 PM, John Snow wrote:
>>> + error_setg_errno(&job->err, -job->ret, "job failed");
>>
>> Kevin specifically asked for me to change this, and I lost it in the
>> shuffle. I'll send a v3 now, since there are enough nits to warrant it,
>> and I think I want to adjust a few things to set up the "part II"
>> portion of this changeset a little more nicely.
>
> But error_setg_errno() uses either strerror() or
> g_win32_error_message(), depending on the host OS. I prefer that over a
> blind strerror(), to be honest.
>
uhh, does it...?
it looks like error_setg_errno is always error_setg_errno_internal,
which always uses strerror... am I misreading?
> In general, it might make sense to introduce a qemu_strerror() (which
> g_strdup()s strerror() on Linux, I presume, so it's compatible with
> Win32); or to allow passing a NULL format to error_setg_errno() so you
> only get the error string.
>
> Max
>
