Alexander, thanks for the review, I've corrected all as you suggested.
Sure I'll control that test-run must be pushed before this patch.
On Thu, May 21, 2020 at 02:29:33PM +0300, Alexander Turenko wrote:
> LGTM.
>
> (But should NOT be pushed to master until I'll push test-run fix from PR
> #211.)
>
> Left a couple of nits, however. See below.
>
> WBR, Alexander Turenko.
>
> > 60f84cbf ('test: use unix sockets for iproto connections')
>
> Nit: We mention commits in the form
> 0000000000000000000000000000000000000000 ('commit message header'): full
> 40-digits hash and the commit message header. You'll need to carry the
> header when will use full commit hash and it is okay.
>
> Personally I think that short hashes lookes better when used inline
> within a text and that our style will recommend them sooner or later.
> But now it is not so.
>
> > diff --git a/test/app-tap/tarantoolctl.test.lua b/test/app-tap/tarantoolctl.test.lua
> > index 4d7059559..dd90c8a25 100755
> > --- a/test/app-tap/tarantoolctl.test.lua
> > +++ b/test/app-tap/tarantoolctl.test.lua
> > @@ -465,12 +465,9 @@ else
> > local remote_path = create_script(dir, 'remote.lua', remote_code)
> > test_run:cmd(("create server remote with script='%s'"):format(remote_path))
> > test_run:cmd("start server remote")
> > - local port = tonumber(
> > - test_run:eval("remote",
> > - "return require('uri').parse(box.cfg.listen).service")[1]
> > - )
> > + local admin_socket = test_run:eval("remote", "return box.cfg.listen")[1]
>
> Nit: Admin port / socket usually refer a console port / socket. I would
> name the variable `listen_url` or `remote_uri`. The former is used
> several times across tests and also is part of `return_listen_uri`
> option name.