Re: [libvirt] [PATCH 2/4] qemu: command: Don't bother reporting errors in smbios formatters

2016-10-13 Thread Peter Krempa
On Thu, Oct 13, 2016 at 17:11:54 -0400, John Ferlan wrote:
> 
> 
> On 10/10/2016 11:51 AM, Peter Krempa wrote:
> > qemuBuildSmbiosBiosStr and qemuBuildSmbiosSystemStr return NULL if
> > there's noting to format on the commandline. Reporting errors from
> 
> nothing
> 
> > buffer creation doesn't make sense since it would be ignored.
> 
> Introduced by 54c0237ccb, so it's been this way a long time...
> 
> > ---
> >  src/qemu/qemu_command.c | 14 --
> >  1 file changed, 14 deletions(-)
> > 
> 
> I suppose since it seems the only legitimate error you'll hit is ENOMEM
> and if you ignore it here, some shortly to be run code is sure to run
> into it, then no big deal...
> 
> Although, one could argue the callers should check/return on error, but
> they'd probably lose that argument.

Not really :). I thought the same when writing the patch as it's fully
possible although extremely unlikely that we'd start a VM with invalid
configuration. Given that we were doing it like this for quite some
while and it would require a more invasive refactoring of the code I
just decided to drop the code altogether.

> ACK for what's here

Thanks, I'll go with this version currently, since it keeps the
semantics present for a rather long time.

Peter


signature.asc
Description: Digital signature
--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list

Re: [libvirt] [PATCH 2/4] qemu: command: Don't bother reporting errors in smbios formatters

2016-10-13 Thread John Ferlan


On 10/10/2016 11:51 AM, Peter Krempa wrote:
> qemuBuildSmbiosBiosStr and qemuBuildSmbiosSystemStr return NULL if
> there's noting to format on the commandline. Reporting errors from

nothing

> buffer creation doesn't make sense since it would be ignored.

Introduced by 54c0237ccb, so it's been this way a long time...

> ---
>  src/qemu/qemu_command.c | 14 --
>  1 file changed, 14 deletions(-)
> 

I suppose since it seems the only legitimate error you'll hit is ENOMEM
and if you ignore it here, some shortly to be run code is sure to run
into it, then no big deal...

Although, one could argue the callers should check/return on error, but
they'd probably lose that argument.

ACK for what's here

John

--
libvir-list mailing list
libvir-list@redhat.com
https://www.redhat.com/mailman/listinfo/libvir-list