[Gt-eos] Launching the GridCF
Brian Bockelman
bbockelm at cse.unl.edu
Mon Oct 30 17:23:45 CET 2017
> On Oct 30, 2017, at 9:07 AM, Mischa Salle <msalle at nikhef.nl> wrote:
>
> Hi Brian, others,
>
> thanks all for all the work done so far!
>
> A few remarks inline...
>
> On Fri, Oct 27, 2017 at 12:39:35PM -0500, Brian Bockelman wrote:
>> Hi all,
>>
>> Frank Scheiner beat me to the punch and put together a GridCF presence
>> on GitHub. Not seeing any negative responses yet, I put together a
>> few scratch webpages and presences (https://gridcf.org
>> <https://gridcf.org/>). Please excuse my enthusiasm.
>>
>> I propose that we should officially launch the Grid Community Forum
>> (GridCF). The GridCF should be an organization to provide broad
>> community-based support for core software packages in grid computing;
>> it is a group of collaborators and not a legal entity. It will not
>> own any copyrights or trademarks on the software it produces;
>> copyright ownership is maintained by the contributor.
> The question I have is whether we only want to use it for community
> supporting a globus-toolkit clone, or whether it should be a 'grid
> community forum' supporting a wide(r) range of grid software? The globus
> toolkit is only a small part of the grid software in use...
> My personal preference is to (at least for the time being) restrict
> ourselves to the former, to keep it simpler and more focussed. It does
> make the name a bit misleading...
I was a bit purposely ambiguous on this.
I would like to "do well" on a GT fork first, but I fully realize there may be other collaborative projects where this is needed. I'd like to have some of the surrounding "infrastructure" there to do wider work, but then narrow it down later.
>
>> Along these lines, I propose we do the following:
>>
>> 1) Propose an initial project management committee. To involve all
>> organizations with a stake in the health of the GridCF, I would
>> like to have one participant in the PMC from each organization that
>> considers themselves a stakeholder.
>> - *However*, after the initial formation of the PMC, I propose all
>> participants be considered individuals. They are certainly
>> allowed (encouraged!) to advocate for their organization's
>> interests, but we don't formally maintain a balance of
>> organizations on the PMC.
>> - That said, I personally think it is strongly in the PMC's interest
>> to maintain a balance across all stakeholders. However, that is
>> up to the PMC and I would be reluctant to codify it in any formal
>> rules.
>> - I propose we should feel free to nominate individuals to represent
>> organizations. I leave it up to the organization to determine
>> their internal process of selecting individuals.
>> - I also propose we let organizations self-identify as wanting to
>> participate and err on the side of inclusion. My initial thinking
>> here is to consider grid organizations (examples including EGI,
>> NorduGrid, OSG, PRACE, XSEDE, etc) but would be fine with a
>> broader scope.
>> - It's hard to manage with too-large groups: I think it would be
>> most productive with 5-10 individuals on the PMC.
> This sounds like a good setup. Perhaps initially 1 per org with a
> replacement, making a total of O(10) but effectively only O(5) ?
> I think some orgs are likely to be only users and not really
> contributors, but they can still have a say in a PMC of course, although
> they might not want to.
Sure. My thinking is it is quite difficult to get rough consensus with a huge group of people -- but if it is too small, you risk not having stakeholders on board.
>
>> 2) I propose we should quickly adopt a simple governance structure. I
>> put together a draft here:
>> https://github.com/gridcf/gridcf.github.io/pull/1
>> <https://github.com/gridcf/gridcf.github.io/pull/1>
>> - Comments are welcome either via mail list or GitHub. Feel free to
>> comment on the issue or post a PR to my branch.
> Looks ok to me. Only thing I wondered about is whether a 3/4 rule is a
> good idea: it will depend on how strongly the against voters would feel
> about their point whether this would work out well or not, but I agree
> we need some way to be able to move forward. Note that if we have 5 PMC
> members of 5 orgs and you have a 3/4 (or 4/5 in that case) voting yes
> and the rest no, we have effectively excluded an org. In such a scenario
> you might want to prefer a unanimous vote and I think it would be rare
> if 5 people cannot come to some form of compromise which is acceptable
> to all.
Yeah - I suspect this will be a rule that gets "touched up" as things start to come together. It's hard to guess exactly where to set the threshold without knowing the composition.
The idea behind this setup was:
1. Requiring a super-majority strongly encourages consensus building if everyone wants to make progress.
2. But not requiring unanimous votes allows progress to be made when things get otherwise stuck.
Indeed - such a setup can allow a single org to be always unhappy and lose every vote. However, I hope that the PMC finds it in their self-interest to avoid this setup.
>
>> 3) With a list of participants and a proposed governance structure, I
>> claim we should vote to officially create the forum with the
>> governance structure posted on the GitHub branch. For the initial
>> vote, I would like the PMC to be unanimous*.
>>
>> Please forgive the level of formality at the very beginning here; I
>> think it would be useful to get this stuff out of the way as soon as
>> possible (as opposed to when the first disagreements hit!). I aim to
>> be less formal once things really get rolling, particularly because
>> I'm not good at these sorts of things.
>>
>> I would also suggest that we try to not nitpick the setup or text:
>> once the forum exists, it is less awkward to fine-tune or change the
>> structure. We *should* make changes necessary to maximize the
>> potential for success.
> I fully agree with that!!
>
>> In terms of timelines:
>> - Between now and Friday, November 3, 2018, we should discuss this
>> proposal and try to build a list of names for the PMC.
>> - Starting Friday, November 3, 2018, we should aim to hold a vote,
>> keeping the vote open until November 7 and posting the results on
>> this list.
>>
>> It's a quick timeline, but I feel it will help us keep from
>> over-thinking it.
>>
>> Once we agree the forum should exist, here's an example of the actions
>> I think we should take rather quickly:
>> - Adopt a code base for the GCT. A few of us will tinker with the
>> repository here (https://github.com/gridcf/gct
>> <https://github.com/gridcf/gct>) in the meantime, but it wouldn't be
>> official until a vote occurs.
>> - Vote on a list of code committers, being as inclusive as reasonable.
>> - Put together a timeline for a first release of the GCT. Given we're
>> working against a January 2018 deadline, I would aim to have the
>> first code release be prior to that and essentially consist of the
>> current fork point from the Globus Toolkit.
> Wouldn't it be better to make the first release just after the last
> globus-toolkit release?
>
> Mischa
>
>
>> Phew - thanks for staying with me! I would like to thank you in
>> advance for enthusiastically contributing to the (hopefully)
>> forthcoming GridCF!
>>
>> Brian
>>
>> * It's a bit tricky to state precisely what unanimity means within a
>> group of people that does not exist in order create their existence.
>> I am hoping that everything I write is as widely acceptable as
>> possible and that we all can come together constructively in order to
>> get things off the ground.
>>
>
>> _______________________________________________
>> Gt-eos mailing list
>> Gt-eos at mailman.egi.eu
>> http://mailman.egi.eu/mailman/listinfo/gt-eos
>
>
> --
> Nikhef Room H155
> Science Park 105 Tel. +31-20-592 5102
> 1098 XG Amsterdam Fax +31-20-592 5155
> The Netherlands Email msalle at nikhef.nl
> __ .. ... _._. .... ._ ... ._ ._.. ._.. .._..
> _______________________________________________
> Gt-eos mailing list
> Gt-eos at mailman.egi.eu
> http://mailman.egi.eu/mailman/listinfo/gt-eos
More information about the Gt-eos
mailing list