Sorry for the late reply, I'm on business trip to US this week and missed this mail at that time.
Maybe we could just start with ARM CI build but just treat it as "optional", and once we are happy with the ARM CI build result, we could count it in. If it goes unstable, ignore it again. Does that make sense? Another possible issue is that contributor may not want to deal with ARM64 as well. Personally I don't deal with Windows machine, because I don't have one, but that's not an only reason. I also hope not to deal with Windows machine cause if something goes wrong specific to Windows, I might have no idea to fix that. So (maybe) dedicated help would be really needed to keep up supporting ARM, but in worst case we can ignore the build, so no big deal. On Sat, Jul 6, 2019 at 4:29 AM Stig Rohde Døssing <[email protected]> wrote: > Jungtaek, does the workflow outlined by Yikun work for you? > > Den tor. 27. jun. 2019 kl. 18.00 skrev Roshan Naik > <[email protected]>: > > > Thanks for volunteering. > > > > > > Sent from Yahoo Mail for iPhone > > > > > > On Wednesday, June 26, 2019, 7:08 PM, Yikun Jiang <[email protected]> > > wrote: > > > > Yes, we will definitely help to fix ARM test failures in PRs, and the > > OpenLab CI will tell us which PR perhaps has ARM compatible problem, if > > it's an easy fix problem, I think authors can fix it by themselves. If > not, > > we will help them to address it. > > > > For the lack of ARM environment, if the developer are interested in > testing > > and debugging their storm patch in ARM environment but don't have ARM > > environment , we can also provide tmp ARM environment to them for > testing. > > > > Regards, > > Yikun > > ---------------------------------------- > > Jiang Yikun（Kero） > > Mail: [email protected] > > > > > > Stig Rohde Døssing <[email protected]> 于2019年6月26日周三 上午1:48写道： > > > > > It sounds pretty low risk for us, if you're volunteering to help fix > any > > > ARM-specific CI failures that may crop up. Will this include helping to > > fix > > > ARM test failures in PRs? Most people are unlikely to have an ARM > > > environment they can use to test, and I'd prefer not to ask > contributors > > to > > > fix ARM test failures themselves. > > > > > > Den tir. 25. jun. 2019 kl. 11.08 skrev Yikun Jiang < > [email protected] > > >: > > > > > > > Sorry for late reply, actually, we have a developer team that willing > > to > > > > work on this, and I'm the owner of storm ARM CI in the OpenLab team. > We > > > not > > > > only want to enable the OpenLab CI, but also want to maintain the arm > > CI > > > > job and fix the CI issue in Storm project. That means if the storm > > > project > > > > has some ARM compatible problem, I will very happy to fix it. > > > > > > > > As I mentioned before, one of the OpenLab goal is to make more open > > > source > > > > software to be more compatible for aarch64 platform. And the Storm > > > project > > > > is one of the most important one in BigData area, so we would like to > > > > propose to work on aarch64 related works in Storm. we plan to start > the > > > > aarch64 related work from to add aarch64 build job for Storm, and our > > > > initial plan as below: > > > > > > > > 1. We first propose enable a periodical job in OpenLab project to > make > > > sure > > > > the storm can compile and build successfully in arm64 env. > > > > 2. We hope can run the OpenLab CI in every pr to ensure that any pr > > after > > > > Step 1 will not break the arm64 build. > > > > 3. We can add more complex test cases on aarch64, like unit tests and > > > > functional tests, step by step. It's a log term works. > > > > > > > > Of course, welcome another developers join to maintain the aarch64 > CI, > > > and > > > > take effort on aarch64 work together. > > > > > > > > Thanks for your attention. > > > > > > > > Regards, > > > > Yikun > > > > ---------------------------------------- > > > > Jiang Yikun（Kero） > > > > Mail: [email protected] > > > > > > > > > > > > Stig Rohde Døssing <[email protected]> 于2019年6月21日周五 上午2:23写道： > > > > > > > > > I guess there is no interest in maintaining ARM compatibility, or > at > > > > least > > > > > no one currently wants to take on the effort. Let's not add the > > Openlab > > > > CI > > > > > then, we can always do it later if someone expresses interest (and > > > > > willingness to maintain). > > > > > > > > > > Den ons. 12. jun. 2019 kl. 17.11 skrev Stig Rohde Døssing < > > > > > [email protected]>: > > > > > > > > > > > Good point, let's see if there's anyone with an ARM environment. > > > > > > > > > > > > Den tir. 11. jun. 2019 kl. 23.07 skrev Jungtaek Lim < > > > [email protected] > > > > >: > > > > > > > > > > > >> I guess the point is not related to open CI build. The point is > > > > whether > > > > > we > > > > > >> really want to support ARM. I'm seeing OpenLab request for other > > > > Apache > > > > > >> projects as well, so I'd rather not treat their request as > > > commitment > > > > of > > > > > >> putting efforts to make builds on ARM green. > > > > > >> > > > > > >> I'm a bit hesitant to add some environment on the support list > > > unless > > > > > >> there're enough engineers willing to work on maintaining. If the > > > work > > > > is > > > > > >> left to existing maintainers, we will be stuck on failing tests > on > > > ARM > > > > > CI > > > > > >> but no environment to play locally. > > > > > >> > > > > > >> So I'd be +1 if there're at least two folks experienced with ARM > > > env. > > > > > >> volunteer to maintain the compatibility part. Otherwise -1 here. > > > > > >> > > > > > >> Thanks, > > > > > >> Jungtaek Lim (HeartSaVioR) > > > > > >> > > > > > >> On Mon, Jun 10, 2019 at 7:27 PM Stig Rohde Døssing < > > > > > >> [email protected]> > > > > > >> wrote: > > > > > >> > > > > > >> > Hi, > > > > > >> > > > > > > >> > An issue was opened suggesting adding ARM CI to the Storm > build, > > > > using > > > > > >> > OpenLab https://issues.apache.org/jira/browse/STORM-3401, > > > > associated > > > > > PR > > > > > >> > here https://github.com/apache/storm/pull/3023 > > > > > >> > > > > > > >> > As far as I can tell, we need to allow OpenLab access to the > > > Github > > > > > PRs, > > > > > >> > similar to how Travis has permission to access our PRs. The > app > > > that > > > > > >> needs > > > > > >> > access is https://github.com/apps/theopenlab-ci. Note that > the > > > app > > > > > will > > > > > >> > have access to the PRs only, not the repo code. > > > > > >> > > > > > > >> > Is anyone opposed to me asking infra to allow OpenLab access > to > > > our > > > > > PRs? > > > > > >> > I'll let this thread sit for a week or so, if there is no > > > opposition > > > > > >> I'll > > > > > >> > ask infra to give the app access. > > > > > >> > > > > > > >> > > > > > >> > > > > > >> -- > > > > > >> Name : Jungtaek Lim > > > > > >> Blog : http://medium.com/@heartsavior > > > > > >> Twitter : http://twitter.com/heartsavior > > > > > >> LinkedIn : http://www.linkedin.com/in/heartsavior > > > > > >> > > > > > > > > > > > > > > > > > > > > > > > > > -- Name : Jungtaek Lim Blog : http://medium.com/@heartsavior Twitter : http://twitter.com/heartsavior LinkedIn : http://www.linkedin.com/in/heartsavior
