[Libguestfs] Having integration tests "test the future" with QEMU

[Markus Armbruster]

"Richard W.M. Jones" <rjones at redhat.com> writes:

> On Wed, Mar 04, 2020 at 03:48:34PM +0100, Markus Armbruster wrote:
>> I understand libguestfs comes with integration tests.  You might be
>> interested in something I'm developing for QEMU 5.0 to permit "testing
>> the future".  I'd like to ensure it is actually useful before I
>> continue.  Let me know what you think.
>> 
>> = Motivation =
>> 
>> When layers above QEMU somehow miss feature deprecation, things continue
>> to work until the deprecated interface goes away, at which point we have
>> an entirely avoidable regression.
>> 
>> We've been trying to get better at communicating feature deprecation to
>> the layers above QEMU.  An obvious first step was systematic
>> documentation.
>> 
>> However, documentation is all too easy to miss.  Even if you don't miss
>> it, you may need to continue using the deprecated feature with old
>> versions of QEMU, and that logic needs to be tested.
>> 
>> I'm working on a way to run QEMU with the deprecated features disabled.
>> This permits "testing the future".
>> 
>> From my cover letter:
>> 
>>     This series extends QMP introspection to cover deprecation.
>>     Additionally, new option -compat lets you configure what to do when
>>     deprecated interfaces get used.  This is intended for testing users of
>>     the management interfaces.  It is experimental.
>> 
>>     -compat deprecated-input=<in-policy> configures what to do when
>>     deprecated input is received.  Available policies:
>> 
>>     * accept: Accept deprecated commands and arguments (default)
>>     * reject: Reject them
>>     * crash: Crash
>> 
>>     -compat deprecated-output=<out-policy> configures what to do when
>>     deprecated output is sent.  Available output policies:
>> 
>>     * accept: Emit deprecated command results and events (default)
>>     * hide: Suppress them
>> 
>>     For now, -compat covers only deprecated syntactic aspects of QMP.  We
>>     may want to extend it to cover semantic aspects, CLI, and experimental
>>     features.
>
> Yes we would add a -compat option when running qemu (only when running
> non-production builds of libguestfs) so that it would fail loudly when
> we use a deprecated feature.  This is the best way I've heard of so
> far to get advanced warning of deprecated features.

Excellent.  I'll let you know when my work is upstream.