Re: [PATCH v1 6/8] iotests: add "compression type" for test output matching

[Eric Blake]

On 2/27/20 4:09 AM, Vladimir Sementsov-Ogievskiy wrote:
> 27.02.2020 13:04, Vladimir Sementsov-Ogievskiy wrote:
> > 27.02.2020 10:29, Denis Plotnikov wrote:
> > > Affected tests: 049, 060, 061, 065, 144, 182, 242, 255
> > >
> > > After adding the compression type feature for qcow2, the compression 
> > > type
> > > is reported on image quering.
> > >
> > > Add the corresponding values of the "compression type" for the tests' 
> > > output
> > > matching.
> >
> > And this and the following patch.
> >
> > Ideally, patch should not break any iotests. This means that all 
> > iotest updates
> > should be merged to the patch which changes their output. Probably, 
> > you can split
> > behavior-changing patch, to reduce iotest-updates per patch, but 
> > anyway, big patch
> > with a lot of iotests updates is better than patch which breaks iotests.
>
> Or we can try to reduce behavior changes in case of zlib:
>
> - keep header small in case of zlib, not adding zero field

No, I don't see the point in doing that.  It makes maintenance harder to 
try to minimize the header, for no gain (older images parse the larger 
header just fine).

> - don't add feature table entry, if compress type is zlib
> - don't report compression type on quering, if it is zlib
>
> - then, all iotests output will be saved. And, then, if we need, we can 
> change
> these theree points one-by-one, updating iotests outputs. But I doubt 
> that we
> really need it, compatible behavior seems good enough.

[by the way, replying to this series is painful, since address@hidden 
is not a valid email address]

--
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org