Re: [PATCH v5 0/5] qcow2: Implement zstd cluster compression method

[Eric Blake <eblake@redhat.com>]

On 3/11/20 11:28 AM, Max Reitz wrote:
> On 11.03.20 08:31, Denis Plotnikov wrote:
>> ping!
>>
>> Is there any other comments/concerns/objections/suggestions according to
>> the series except the minor ones from Alberto and Vladimir?
>> If not, please, let me know, so I can resend the series with the minor
>> changes for applying to the corresponding branch.
> 
> Sounds good to me.
> 
> I’d like to note that most iotests that seem to do something with
> compression (i.e., where grep finds a 'compress' somewhere; 013, 014,
> 023, 042, 046, 053, 055, ...) pass with -o compression_type=zstd, too.
> 060 hangs somewhere.  112 complains about v2 incompatibility; and 214
> relies on intricacies of zlib, I think.  So that looks good, too.
> 
> Well, one thing I did have to fix for this to work is to quote
> everything in common.pattern that tries to echo something with brackets
> (e.g. “Clusters to be compressed [1]”).  I don’t quite know why the
> brackets suddenly disappear when I run the tests with -o
> compression_type, who knows.

That sounds like you have a file named '1' in the directory where the 
echo was run.  Unquoted [1] is a shell glob that expands to '1' if that 
file exists, otherwise remains unexpanded as '[1]'.

>  But putting quotes around the echo
> arguments fixes it, so...

Yes.  Figuring out why a file named '1' is being created is also a 
useful exercise, but quoting any time you have [ that you want output is 
already a good idea.

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