But that means changing the input file from using the default engine, to using "qemu" explicitly.
I was thinking that keeping the input xml unchanged gets a stronger test, since it also checks that default uses qemu.
Also since most users will still not use the new engine property, so keeping the input xml unchanged also tests the more common workload.

-----"Peter Krempa" <pkrempa@redhat.com> wrote: -----


>To: "Or Ozeri" <ORO@il.ibm.com>
>From: "Peter Krempa" <pkrempa@redhat.com>
>Date: 10/07/2021 03:26PM
>Cc: libvir-list@redhat.com, idryomov@gmail.com,
>to.my.trociny@gmail.com, "Danny Harnik" <DANNYH@il.ibm.com>
>Subject: [EXTERNAL] Re: [PATCH v3 4/5] conf: add encryption engine
>property
>
>On Thu, Oct 07, 2021 at 11:31:42 +0000, Or Ozeri wrote:
>> -----"Peter Krempa" <[1]pkrempa@redhat.com> wrote: -----
>> >> diff --git a/tests/qemuxml2xmloutdata/disk-nvme.xml
>> >b/tests/qemuxml2xmloutdata/disk-nvme.xml
>> >> deleted file mode 120000
>> >> index ea9eb267ac..0000000000
>> >> --- a/tests/qemuxml2xmloutdata/disk-nvme.xml
>> >> +++ /dev/null
>> >> @@ -1 +0,0 @@
>> >> -../qemuxml2argvdata/disk-nvme.xml
>> >
>> >Files that were originally symlinks ...
>> >
>> >> \ No newline at end of file
>> >> diff --git a/tests/qemuxml2xmloutdata/disk-nvme.xml
>> >b/tests/qemuxml2xmloutdata/disk-nvme.xml
>> >> new file mode 100644
>> >> index 0000000000..9a5fafce7d
>> >> --- /dev/null
>> >> +++ b/tests/qemuxml2xmloutdata/disk-nvme.xml
>> >
>> >... must not be expanded to full files. That is a drawback of
>using
>> >sed
>> >-i to do the conversion.
>> Should I create a new file
>> e.g. tests/qemuxml2xmloutdata/disk-nvme-content.xml
>> and point the disk-nvme.xml symlink to it?
>
>No just keep symlinks as symlinks. You are not changing anything that
>would warant a change to full output file. Namely if the output file
>is
>a symlink to the input file, then the change to the input file should
>be
>sufficient as the output file is the same one.
>
>