Using gerrit or github for review?

[Heiko Schocher]

Hi all,

Am 13.07.2020 um 23:03 schrieb Tom Rini:
> On Mon, Jul 13, 2020 at 10:12:32PM +0200, Wolfgang Denk wrote:
>> Dear Simon,
>>
>> In message <CAPnjgZ2OnxotZ3gGErO0KfFJVA6_jSY6xO_Kg63Ts86qTw4KBg at mail.gmail.com> you wrote:
>>>
>>> I have used various tools and I'm wondering whether having another
>>> option might have some benefits in terms of productivity, automation
>>> and accessibility. Just as one example, if people pushed patches to
>>> github / gitlab then we could 1) check out the branch and try it, 2)
>>> have test automation attached, 3) use a UI for review.
>>
>> As for automation, you know that Heiko has been using tbot for years
>> for automatic patch series download from patchwork, checking and
>> testing (checkpatch, build, install, boot)?
> 
> Perhaps this is the time we need to figure out what's missing from tbot
> automation and related tooling and get things integrated well between
> tbot, test/py/ testing we have and other ad-hoc tooling.

Yes, but downloading patches from patchwork is missing in the new tbot
implementation yet... but it worked for me very well ... I think,
I made a video on youtube... ah yes:

https://www.youtube.com/watch?v=ZwUA0QNDnP4&t=55s

but without any comment, so not very good to follow the log, may with
lower speed...

The video starts where I download the patches from my Patchwork ToDo
list and apply them to current U-Boot code, build and install U-Boot
(here for the smartweb board) and do tests on U-Boots commandline ...
If finished, pushed the result to a db...

The new implementation misses currently only the step "download
patches from Patchwork ToDo list" [1] which should take not much effort
to reimplement. The new tbot already can handle git trees [2] and
so can apply patches to it.

Of course tbot can configure/build/install U-Boot [3] and call
test.py [4] and push results to a DB [5].

For a demo I do this once a day currently here:

http://xeidos.ddns.net/ubtestresults/home

for at least 4 boards (I am happy to get more testreports! You
do not necessarily need tbot for it ...)

Also, of course, tbot can do a "git bisect" [6] so may you can find
out, which patch breaks ...

As tbot is a commandline tool, of course you can start tbot from
gitlab CI runner, jenkins, buildbot,...

I tried to start tbot from a travis build, but did not find (yet) a way
to connect from the travis builder over ssh to a lab, where the boards
under test are located ...

Note:
The main purpose of tbot is to help a developer while his daily work,
and if you use it for your daily work, you automatically have a setup
ready for starting from a CI ... I personally use the minimal way, and
call tbot from a small python script (cron job like) [7]

bye,
Heiko

[1] old tbot "download patches from patchwork ToDo list" testcase:
https://github.com/hsdenx/tbot/blob/master/src/tc/linux/tc_workfd_get_patchwork_number_list.py

[2] http://tbot.tools/modules/tc.html#git

[3] http://tbot.tools/modules/tc.html#u-boot

[4] http://tbot.tools/modules/tc.html#u-boot-test-py

[5] https://github.com/EmbLux-Kft/tbot/blob/devel/generators/push-testresult.py
     Not mainline yet, experimental state!

[6] http://tbot.tools/modules/tc.html#tbot.tc.git.GitRepository.bisect

[7] https://github.com/EmbLux-Kft/tbot-tbot2go/blob/devel/minimal-ci.py
     config:
     https://github.com/EmbLux-Kft/tbot-tbot2go/blob/devel/minimal-ci.json
-- 
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: +49-8142-66989-52   Fax: +49-8142-66989-80   Email: hs at denx.de