[CRIU] CRIU and GO

[Radostin Stoyanov]

On 04/10/2018 23:11, Andrei Vagin wrote:
> On Thu, Oct 4, 2018 at 12:40 PM Adrian Reber <adrian at lisas.de> wrote:
>> On Wed, Oct 03, 2018 at 05:37:25PM +0200, Adrian Reber wrote:
>> [...]
>>>>>>> What do you say? Would it make sense to have an independent go binding
>>>>>>> for CRIU? Once it works we could start to also use it in runc...
>>>>> The main idea was to have a canonical location for the go bindings which
>>>>> is easy to detect and switch runc to using it once it exists. I would
>>>>> like to use it in podman right now and the vendor/github.com tree is
>>>>> around 20MB of which 5MB is CRIU. It basically is only a tool problem
>>>>> (vndr), but it probably is more difficult to change the tool, then
>>>>> provide CRIU go bindings separately.
>>>> Ok. All these sound reasonable. I don't have objections about a new repo.
>>> OK, let me prepare something and I will post it here for review.
>> I am looking at the phaul GO implementation and I think it would make
>> sense to include it into the CRIU GO bindings repository. Because it is
>> kind of the reference implementation/user for the go bindings right now
>> and would be a good test case. Would you rather keep it in the criu
>> repository? If necessary we could pull in the CRIU GO bindings
>> repository in CRIU during travis tests.
> If we are going to move the go library into a separate repo, I would
> say that we have to create a separate repo for the phaul library too.
I like the idea of moving the go library and phaul into a separate repo.

What about crit and pycriu, can we move them into a separate repo as well?
Then we can make them available on PyPI. This will make it possible to be
installed with pip (e.g. pip install pycriu) and referenced as a dependency
from other python applications.

Or do we need them in the repo because of ZDTM?

Just a thought.

Radostin