[P4-design] toward P4_16 closure

Anirudh Sivaraman anirudh at csail.mit.edu
Sun Oct 23 16:21:43 EDT 2016


On Sun, Oct 23, 2016 at 3:17 PM, Changhoon Kim
<chang at barefootnetworks.com> wrote:
> All,
>
> If you want to discuss certain git issues tomorrow at the design meeting,
> please submit your list by tomorrow morning.
>

I would like to discuss issue 48 (the concurrency model). As Mihai's
mail suggested, I think this issue is solved, and several people have
contributed to the github issue. At the same time, it's an important
structural change, and I would like to address any concerns people on
the group may have with it.

I also think issue 81 (local externs) should be discussed, but it's
lower priority and can likely be settled over email.


> -- Chang
>
> On Thu, Oct 20, 2016 at 10:51 AM, Mihai Budiu <mbudiu at vmware.com> wrote:
>>
>> If anyone would like to veto closing any of the issues marked as “should
>> be solved”, please make your voice heard. Otherwise we will assume that the
>> filer of the issue is the one who determines the actual consensus. I believe
>> we have closed several issues in a similar fashion already.
>>
>>
>>
>> Mihai
>>
>>
>>
>> From: Ben Pfaff
>> Sent: Thursday, October 20, 2016 10:49 AM
>>
>>
>> To: Mihai Budiu <mbudiu at vmware.com>; Changhoon Kim
>> <chang at barefootnetworks.com>; p4-design at lists.p4.org
>> Subject: Re: [P4-design] toward P4_16 closure
>>
>>
>>
>> OK, I thought that we were only supposed to close issues after coming to
>> consensus.  If it's OK to close one's own issues then that's fine with me
>> too.
>>
>> ________________________________
>>
>> From: Mihai Budiu
>> Sent: Thursday, October 20, 2016 10:25:18 AM
>> To: Ben Pfaff; Changhoon Kim; p4-design at lists.p4.org
>> Subject: RE: [P4-design] toward P4_16 closure
>>
>>
>>
>> I think that ideally the people who have filed the issues should be able
>> to decide (unless there is a veto for solving from someone else). I am
>> always available to offer clarifications on why I expect each of them to
>> have been solved – we can use personal email for such pointed questions if
>> we don’t want to pollute the mailing list. If the original filer cannot
>> decide we should take a poll.
>>
>>
>>
>> Mihai
>>
>>
>>
>> From: Ben Pfaff
>> Sent: Thursday, October 20, 2016 10:21 AM
>> To: Mihai Budiu <mbudiu at vmware.com>; Changhoon Kim
>> <chang at barefootnetworks.com>; p4-design at lists.p4.org
>> Subject: Re: [P4-design] toward P4_16 closure
>>
>>
>>
>> Would it be worthwhile to just take a poll on the "should be solved" ones
>> so that we can close them without spending valuable in-person time?
>>
>> ________________________________
>>
>> From: P4-design <p4-design-bounces at lists.p4.org> on behalf of Mihai Budiu
>> <mbudiu at vmware.com>
>> Sent: Thursday, October 20, 2016 10:14:24 AM
>> To: Changhoon Kim; p4-design at lists.p4.org
>> Subject: Re: [P4-design] toward P4_16 closure
>>
>>
>>
>> I would like to discuss the following issues at the next meeting; most of
>> them are relatively small and confined.
>>
>>
>>
>> Mihai
>>
>>
>>
>> https://github.com/p4lang/p4-spec/issues/86 (Remove some annotations from
>> the spec)
>>
>> https://github.com/p4lang/p4-spec/issues/78 (prototypes) - should be
>> solved by new commits
>>
>> https://github.com/p4lang/p4-spec/issues/77 (namespaces and visibility) -
>> should be solved
>>
>> https://github.com/p4lang/p4-spec/issues/63 (Checksum definitions) -
>> should be solved
>>
>> https://github.com/p4lang/p4-spec/issues/59 (tuple types) - should be
>> solved
>>
>> https://github.com/p4lang/p4-spec/issues/54 (reserved annotation names)
>>
>> https://github.com/p4lang/p4-spec/issues/48 (concurrency model) - should
>> be solved
>>
>> https://github.com/p4lang/p4-spec/issues/43 (copy-in copy-out) - should be
>> solved
>>
>> https://github.com/p4lang/p4-spec/issues/60 (tuples and calling
>> conventions)
>>
>> https://github.com/p4lang/p4-spec/issues/58 (switch vs. select)
>>
>> https://github.com/p4lang/p4-spec/issues/55 (parser and control cross
>> invocations)
>>
>>
>>
>> From: P4-design [mailto:p4-design-bounces at lists.p4.org] On Behalf Of
>> Changhoon Kim
>> Sent: Wednesday, October 12, 2016 11:19 PM
>> To: p4-design at lists.p4.org
>> Cc: Amin Vahdat <vahdat at google.com>
>> Subject: [P4-design] toward P4_16 closure
>>
>>
>>
>> Hi P4 designers,
>>
>>
>>
>> As we're all working busily together to release P4_16 in time, I'd like to
>> propose a few ideas to facilitate our on-going review process.
>>
>>
>>
>> - Let's freeze the feature-request list for P4_16. Starting this Friday
>> (Oct/14), I will tag all new git issues for p4lang/p4-spec with the "Post
>> P4_16" milestone. The only exception will be bug reports for the prototype
>> P4_16 compiler frontend (p4lang/p4c).
>>
>>
>>
>> - We originally planned to close all the issues and produce a pre-release
>> version of the P4_16 spec by Oct/24, but it seems we need a few more
>> meetings to be able to finish up our work. I've scheduled the following
>> face-to-face meetings. Note each meeting will be 2hrs long (as opposed to
>> 1.5hrs) starting with the next one. Let's try to finish all technical
>> discussions by Nov/7 and leave Nov/21 for buffer.
>>
>> ?         Oct/24, Monday, 1:30 - 3:30pm, Gates 300
>>
>> ?         Nov/7, Monday, 1:30 - 3:30pm, Gates 300
>>
>> ?         Nov/21, Monday, 1:30 - 3:30pm, Gates 104
>>
>> - Starting next meeting, let's employ a "close-by-default" policy.
>> Representatives who want to discuss particular git issues must explicitly
>> propose those issues by Friday the week before the meeting. Issues that are
>> not explicitly proposed for discussion will be automatically resolved and
>> won't be considered further for P4_16.
>>
>>
>>
>> Thanks a lot for your contributions!
>>
>>
>>
>> -- Chang
>
>
>
> _______________________________________________
> P4-design mailing list
> P4-design at lists.p4.org
> http://lists.p4.org/mailman/listinfo/p4-design_lists.p4.org
>
>




More information about the P4-design mailing list