[Mpi-forum] Time To Open Source MPI

Wesley Bland work at wesbland.com
Mon Feb 3 14:11:40 CST 2020


We’ll be moving from our existing Jenkins server to GitHub Actions. The benefit is that this runs on GitHub’s servers and is easy to update when we need to update the environment (e.g. requiring a newer version of Python like we do now). I’ve been able to test this in a side repository and it works well for what we need. It’s also free for public repositories. 

> On Feb 3, 2020, at 2:06 PM, Joseph Schuchart via mpi-forum <mpi-forum at lists.mpi-forum.org> wrote:
> 
> Wesley,
> 
> I do not have a strong opinion on this question. However, you keep mentioning that the testing infrastructure is broken. I wasn't at the last meeting so I must have missed the discussion there. How would moving to a repo restore the testing infrastucture?
> 
> Cheers
> Joseph
> 
>> On 2/3/20 8:59 PM, Wesley Bland via mpi-forum wrote:
>> I really want to emphasize how little work this should be for each person. For people with lots of PRs (like you), I estimate it’ll be less than 15 minutes before the meeting and less than 15 minutes afterward. For people with no open pull requests, it’s 0 minutes. If people are having trouble, I can do “almost” everything for you (other than clicking the delete and fork buttons). I’m asking for help here because I’d rather make a small amount for a handful of people instead of a lot of work for just me. This is also one of the reasons I want to do this in person, so I can help those having trouble.
>> The reason we want to do it now is because this only became a problem at the last meeting when we broke the testing infrastructure. We could keep pushing this back, but in the meantime, we won’t be getting the benefit of that testing that we’ve relied on for a while to help make sure that Bill has an easier time of merging our pull requests (which will be even more important as we start to finish up these votes).
>> We didn’t make this decision just to give people more work and I spent a fair amount of time making this as painless for everyone as possible. However, unlike some of the other transitions that we’ve made in the past(e.g. SVN -> Git), we also didn’t have the benefit of a long run-up to make this change.
>> Thanks,
>> Wesley
>>>> On Feb 3, 2020, at 1:36 PM, HOLMES Daniel <d.holmes at epcc.ed.ac.uk <mailto:d.holmes at epcc.ed.ac.uk>> wrote:
>>> 
>>> Hi Wesley,
>>> 
>>>> This change really won’t be as painful as it sounds
>>> 
>>> Then why do I have to do a bunch of unexpected/unscheduled backup and book-keeping work before an arbitrarily chosen deadline? I’m not sure that springing this surprise on folks is the wisest course of action.
>>> 
>>> Cheers,
>>> Dan.
>>>>>> Dr Daniel Holmes PhD
>>> Architect (HPC Research)
>>> d.holmes at epcc.ed.ac.uk <mailto:d.holmes at epcc.ed.ac.uk>
>>> Phone: +44 (0) 131 651 3465
>>> Mobile: +44 (0) 7940 524 088
>>> Address: Room 2.09, Bayes Centre, 47 Potterrow, Central Area, Edinburgh, EH8 9BT
>>>>>> The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336.
>>>>>> 
>>>> On 3 Feb 2020, at 16:02, Wesley Bland via mpi-forum <mpi-forum at lists.mpi-forum.org <mailto:mpi-forum at lists.mpi-forum.org>> wrote:
>>>> 
>>>> Because in the meantime, our testing remains broken and we don’t have another solution to fix it at the moment. We really want to have that testing fix.
>>>> 
>>>> This change really won’t be as painful as it sounds. There’s almost no Git commands required. No rebasing/merging/etc.
>>>> 
>>>>> On Feb 3, 2020, at 9:57 AM, Rolf Rabenseifner <rabenseifner at hlrs.de <mailto:rabenseifner at hlrs.de>> wrote:
>>>>> 
>>>>> Why aren't we doing this when we finished 4.0, i.e. merged all the voted in
>>>>> pull requests into the standard?
>>>>> 
>>>>> ----- Original Message -----
>>>>>> From: "Main MPI Forum mailing list" <mpi-forum at lists.mpi-forum.org <mailto:mpi-forum at lists.mpi-forum.org>>
>>>>>> To: "Main MPI Forum mailing list" <mpi-forum at lists.mpi-forum.org <mailto:mpi-forum at lists.mpi-forum.org>>
>>>>>> Cc: "Wesley Bland" <work at wesbland.com <mailto:work at wesbland.com>>
>>>>>> Sent: Monday, February 3, 2020 4:38:48 PM
>>>>>> Subject: [Mpi-forum] Time To Open Source MPI
>>>>> 
>>>>>> Hi folks,
>>>>>> 
>>>>>> ** Look at the numbered list for your action items before 2020-02-18! **
>>>>>> 
>>>>>> Since we moved from Subversion to GitHub, there’s been a back and forth on
>>>>>> whether the MPI Standard Source code should be open source or closed source. As
>>>>>> you know (since most of you have access to it, we decided to keep the standard
>>>>>> closed for a few reasons (avoiding rogue copies of the Standard was a big one),
>>>>>> but recently, the scales have tipped back toward open sourcing and after
>>>>>> talking to Bill and Martin some more, we think it’s finally time to open up the
>>>>>> MPI Standard repository. Specifically, these are the issues that get addressed
>>>>>> by moving to a public repository:
>>>>>> 
>>>>>> * Participation to the forum has a barrier to entry (i.e. me)
>>>>>> * Our current automated testing system is broken and will not be fixed
>>>>>> * We rely on special dispensation from GitHub for free, private repositories
>>>>>> 
>>>>>> I’ll be discussing this in more detail in Portland, but I want to make folks
>>>>>> aware that this is coming now because it will have some impact on people who
>>>>>> are actively working on text proposals. The mitigations here should be pretty
>>>>>> straightforward and I have both slides (for a high level) and a wiki page (for
>>>>>> details) on what’s happening and what you need to do about it.
>>>>>> 
>>>>>> The short version for the purpose of this email and your action item leading up
>>>>>> the forum next week is to do two thing s:
>>>>>> 
>>>>>> 1. Back up your branches.
>>>>>> 2. Make a note for each of your open pull requests on where the are and what
>>>>>> issues they’re attached to.
>>>>>> 
>>>>>> Detailed notes and the slides I’ll present in Portland on how to do this are
>>>>>> here: [
>>>>>> https://github.com/mpi-forum/mpi-issues/wiki/Migrate-to-Public-Repository |
>>>>>> https://github.com/mpi-forum/mpi-issues/wiki/Migrate-to-Public-Repository ]
>>>>>> 
>>>>>> The plan is for me to flip this switch DURING THE FEBRUARY MEETING! This should
>>>>>> not disrupt any readings for the meeting itself as none of the pull requests
>>>>>> will actually be deleted, but they will all become stale and we’ll need to
>>>>>> re-open new ones. The main benefit of doing this during the meeting is that
>>>>>> if/when anyone has trouble, I’ll be there in person to provide live help to
>>>>>> anyone else that’s there (sorry, Marc-Andre, you’ll need to follow my slides
>>>>>> remotely \uD83D\uDE00).
>>>>>> 
>>>>>> Please email me if you have questions or concerns. If there’s enough demand, I’d
>>>>>> even be willing to do a short Virtual Meeting on this on Wednesday.
>>>>>> 
>>>>>> Thanks,
>>>>>> Wesley
>>>>>> 
>>>>>> _______________________________________________
>>>>>> mpi-forum mailing list
>>>>>> mpi-forum at lists.mpi-forum.org <mailto:mpi-forum at lists.mpi-forum.org>
>>>>>> https://lists.mpi-forum.org/mailman/listinfo/mpi-forum
>>>>> 
>>>>> -- 
>>>>> Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner at hlrs.de <mailto:rabenseifner at hlrs.de> .
>>>>> High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530 .
>>>>> University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832 .
>>>>> Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner <http://www.hlrs.de/people/rabenseifner> .
>>>>> Nobelstr. 19, D-70550 Stuttgart, Germany . . . . (Office: Room 1.307) .
>>>> _______________________________________________
>>>> mpi-forum mailing list
>>>> mpi-forum at lists.mpi-forum.org <mailto:mpi-forum at lists.mpi-forum.org>
>>>> https://lists.mpi-forum.org/mailman/listinfo/mpi-forum
>>> 
>> _______________________________________________
>> mpi-forum mailing list
>> mpi-forum at lists.mpi-forum.org
>> https://lists.mpi-forum.org/mailman/listinfo/mpi-forum
> _______________________________________________
> mpi-forum mailing list
> mpi-forum at lists.mpi-forum.org
> https://lists.mpi-forum.org/mailman/listinfo/mpi-forum


More information about the mpi-forum mailing list