Re: [edk2-announce] Community Meeting Minutes

[Rebecca Cran <rebecca@bluestop.org>]

As a casual contributor, for me the biggest complaint I have is how busy 
the mailing list gets. I don't think a new 'announce' list is what's 
needed, perhaps a 'reviews' or 'discussion' list to split out 
discussions (from anyone) from day-to-day patches? Also, I'd be anxious 
about jumping to a new service like groups.io: most open source 
developers understand plain email, and personally I'd like that to stay. 
For example FreeBSD set up web forums, but most contributors continue to 
use the existing mailman based lists, and I suspect tend to forget the 
web interface exists.


One thing I feel that's missing from the current Github-based 
infrastructure of the web site and wiki is that as far as I know there's 
no API documentation built regularly, or automated builds etc. I'm 
hosting the API documentation at e.g. 
https://code.bluestop.org/edk2/docs/master/ . Also, one thing a review 
system like Gerrit, Github, Phabricator, Review Board etc. would give us 
is the ability to run tests (lint, build/run OVMF etc.) against patches 
and have it comment on the review about its status to give committers 
more confidence in it.


-- 

Rebecca Cran


On 2/14/19 12:07 PM, Jeremiah Cox via edk2-devel wrote:
> Hi Ard,
> My apologies as no insult was intended.  The suggestion is that, similar to today, folks encountering difficulties with our systems could reach out to the TianoCore discussion venue (our mailing list or groups.io), and our friendly community members (we have many) will surely assist them.
>
> You are correct that my focus is not casual contributors, rather I want to encourage a large number of UEFI developers who are currently closed to stop their fork-modify-ship model, which is inefficient to service, go open to share their learnings, get current, stay current, upstream their changes (where it makes sense to the community), but not throw garbage over the wall.   I think there is some value in this endeavor.
>
> Kind Regards,
> Jeremiah
>
> ________________________________
> From: Ard Biesheuvel <ard.biesheuvel@linaro.org>
> Sent: Friday, February 8, 2019 5:58 AM
> To: Jeremiah Cox
> Cc: stephano; edk2-devel@lists.01.org; Kinney, Michael D; Laszlo Ersek
> Subject: Re: [edk2] [edk2-announce] Community Meeting Minutes
>
> On Thu, 7 Feb 2019 at 18:52, Jeremiah Cox via edk2-devel
> <edk2-devel@lists.01.org> wrote:
>> Apologies on the late reply, I was on vacation for several weeks and just got back to this.
>>
>> Regarding "Patch Review System Evaluation", on the call, I disagreed with your conclusion, but that note is not captured below.  My reading of the email and call discussions, I did not hear our community reject GitHub, rather observations that it was not "perfect", that it does not transparently interact with folks who prefer mailing list patch systems, but it would be acceptable to try.  On the call you mentioned a second justification for staying with the mailing list system, that GitHub (really all modern patch management systems) exclude folks who have limited internet connectivity.  I hypothesize that this theoretical group of Tianocore contributors would be a very small group of folks.  Should our community optimize our systems for a very small, theoretical group, penalizing the overwhelming majority?  I would propose that we try a modern patch management system, GitHub had the best reviews in my reading, and folks with limited internet connectivity find a friend to act as a go between translating their email diffs into GitHub PRs.
> I find this unnecessarily condescending. Finding a friend, seriously?
>
> Very serious concerns have been raised about the lack of transparency
> with the various systems, and the fact that I am able to consult my
> own local copy of the entire review history, including all email
> exchanges is a very important aspect of the current model to me, as
> opposed to GitHub deciding what is important enough to keep and what
> is not. In an open source project, the code base is *not* the HEAD
> commit, it is the entire repository, including history, and logged
> email threads with technical discussions, since they are usually not
> captured in other ways.
>
> The push to GitHub is being sold to us as a way to attract more
> contributors, but it seems to me (and I have stated this multiple
> times) that the mailing list is not the steep part of the learning
> curve when contributing to TianoCore. So is this really about getting
> outsiders to contribute to Tianocore? Or is it about reducing the
> impedance mismatch with what internal teams at MicroSoft (and Intel?)
> are doing, which probably already went through the learning curve when
> it comes to other aspects of Tianocore.
>
>  From a high level, it might seem that using a mailing list is the
> impediment here. But in reality, contributing to open source in
> general is not about whether you use GitHub or a mailing list to throw
> your stuff over the fence. It is about collaborating with the
> community to find common ground between the various sometimes
> conflicting interests, and permitting your engineers to engage with
> the community.
>
> Tianocore has always been a rather peculiar open source project, since
> it wasn't more than just that, i.e., openly available source code.
> This has been changing for the better during the time I have been
> involved, and we have worked very hard with the Intel firmware team
> and other contributors to collaborate better on the mailing list.
>
> To summarize my concern here: it seems that this push is not about
> making it easier for contributors that already know how to do open
> source collaboration to contribute to Tianocore, it is about making it
> easier for currently closed code to be contributed to Tianocore by
> teams who have no prior experience with open source.
>
> Apologies if I have the wrong end of the stick here. If not, why don't
> we consult a few casual contributors (which should be easy to find on
> the mailing list) and ask them what their biggest issues were with
> contributing to Tianocore?
>
>
>
>
>
>
>> -----Original Message-----
>> From: edk2-devel <edk2-devel-bounces@lists.01.org> On Behalf Of stephano
>> Sent: Friday, January 11, 2019 11:27 AM
>> To: edk2-devel@lists.01.org
>> Cc: Kinney, Michael D <michael.d.kinney@intel.com>; Laszlo Ersek <lersek@redhat.com>
>> Subject: [edk2] [edk2-announce] Community Meeting Minutes
>>
>> An HTML version is available here:
>> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.tianocore.org%2Fminutes%2FCommunity-2019-01.html&amp;data=02%7C01%7Cjerecox%40microsoft.com%7Ce1986594f0094058f09208d68dcd9b0c%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636852311581785508&amp;sdata=EVNgiM90x5nka9boa%2BVsCPVEJjib%2BfcDpQFLJ5m27cs%3D&amp;reserved=0
>>
>> Community Updates
>> -----------------
>> Several conferences are coming up that we will be attending.
>>
>> FOSDEM 2019
>> Stephano will be giving a talk with Alexander Graf (SUSE) on UEFI usage on the UP Squared board and Beagle Bone Black.
>>
>> More info on FOSDEM here:
>> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Ffosdem.org%2F2019%2F&amp;data=02%7C01%7Cjerecox%40microsoft.com%7Ce1986594f0094058f09208d68dcd9b0c%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636852311581795501&amp;sdata=1weJ37WVTOJP4Et%2BgUJqF2KGIfV5g6IlGXEV8n0Lelw%3D&amp;reserved=0
>>
>> Info on the talk here:
>> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Ffosdem.org%2F2019%2Fschedule%2Fevent%2Fuefi_boot_for_mere_mortals%2F&amp;data=02%7C01%7Cjerecox%40microsoft.com%7Ce1986594f0094058f09208d68dcd9b0c%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636852311581795501&amp;sdata=BHkTSCSGQ71rh1G2zr%2FTFtxnzvUXK47vHES7hs0Cvh4%3D&amp;reserved=0
>>
>> Open Compute Project Global Summit
>> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.opencompute.org%2Fsummit%2Fglobal-summit&amp;data=02%7C01%7Cjerecox%40microsoft.com%7Ce1986594f0094058f09208d68dcd9b0c%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636852311581795501&amp;sdata=8Wer0jAgTX2pMeHddxcNdCXmAblGy5pVTfsotl6n1xE%3D&amp;reserved=0
>>
>> No TianoCore talks were accepted for this event, however Stephano will be talking about CHIPSEC.
>> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fsched.co%2FJinT&amp;data=02%7C01%7Cjerecox%40microsoft.com%7Ce1986594f0094058f09208d68dcd9b0c%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636852311581795501&amp;sdata=l3DULTiWsTfbxEoupZ1EbM6SJ2bsHFqK1rVIdl6oolY%3D&amp;reserved=0
>>
>> Other Upcoming Conferences
>> Linuxfest NW
>> PyCon
>> Redhat Summit
>> RustConf
>>
>> Rust
>> ----
>> Stephano is working with some folks from the Open Source Technology Center at Intel regarding their desire to get Rust ported to EDK2. While there are many proof of concepts out there, the first step for adoption would be to integrate the Rust infrastructure into our build system, and create a simple "hello world" app. The goal would be to provide a modern language with better memory safety for writing modules and drivers. Our hope is that the availability of this language would encourage outside contribution and support from a vibrant and well established open source community.
>>
>> Github Discussions Evaluation, Groups.io, Microsoft Teams
>> ---------------------------------------------------------
>> During our December community meeting, we talked about trying out "GitHub Discussions" as a basis for communication that might be better than our current mailing list situation. The main issues with the mailing list today are:
>>
>> 1. Attachments are not allowed.
>> 2. Email addresses cannot be "white listed" (If you are not subscribed your emails are simply discarded by the server).
>>
>> In order to save us some time, Stephano reviewed GitHub discussions using 3 GitHub user accounts, and found the following shortcomings:
>>
>> 1. No support for uploading documents, only images 2. No way to archive discussions outside GitHub[1] 3. Any comment can be edited by any member 4. Discussions are not threaded
>>
>> [1] Email notification archiving is possible, but this means we'd have to keep a mailing list log of our conversations. At that point, why not just use email?
>>
>> That last one is particularly difficult to work around. Every comment is added to the bottom of the list. If some small group of developers (out of many) start having a “sub discussion”, their replies will not be separate from the main thread. There’s no way to distinguish and visually “collapse” a sub thread, so one is forced to view the discussion as a whole. It would seem that the "discussion feature" was intended for small, single threaded discussions. This will not work for larger complex system design discussions.
>>
>> Also, the ability to edit comments is perplexing. Any member can edit any comment, and delete any of their comments or edits. No email notifications are provided for these actions, so there may be no document trail for parts of the conversation. This system seems quite inadequate for serious development discussions and is clearly meant for a more "chat" style of communication on smaller teams. Comments and questions regarding "GitHub Discussions" are still welcomed, but Stephano recommends we move forward with trying out different systems with more robust feature sets.
>>
>> It was agreed that we will evaluate Groups.io next to see if that is a better fit for our needs. Stephano will setup accounts as needed and do some preliminary testing. If that goes well he will initiate discussions on "Line Endings" as well as "Use of C Standard Types".
>>
>> Microsoft Teams was also brought up as a possible solution. If Groups.io fails to provide a good platform for us, we will look into Teams. The main barrier to entry there may be the cost. We have found that many of the software options we have been evaluating have this cost barrier to entry. We need to decide if this is truly a "no-go" issue for using software as a community. If TianoCore was an organization that had non-profit status, it might be easier for us to get non-profit discounts on software like this. Stephano will bring this up at the Steward's Meeting next week.
>>
>> Patch Review System Evaluation
>> ------------------------------
>> After evaluating Github, Gitlab, and Phabricator, we will be remaining with the mailing list for now. Github did prove a possible "2nd runner up" (albeit distant). Also, Stephano / Nate from Intel will be reviewing Gerrit use with a report being sent back to the community sometime next week.
>>
>> Community CI Environment
>> ------------------------
>> Azure DevOps, Cirrus CI, Jenkins, Avacado We will begin evaluation of possible community test frameworks. This again brings up the question of how we would fund such an effort, and Stephano will bring this up at the Steward's meeting. It is important to remember that our supported environments are Linux, Windows, and macOS.
>> We have compilers that are considered "supported" and those combinations should have proper coverage. Also, we do not want to use multiple CI environments, so the solution we choose should support all use cases.
>> There are several CI options that are "Free for open source" but they limit the size / number of CI agents, with pricing tiers for larger sized builds. The cost of a CI infrastructure will be dependent on the number of patches we need to send through the service, and what kind of response is required. Stephano will work with Philippe on Avacado, the folks at MS will evaluate possible use of Azure DevOps (again, possibly limited by the fact that we are not a non-profit), and volunteers are still required to test Cirrus and Jenkins.
>>
>> Public Design / Bug Scrub Meetings
>> ----------------------------------
>> We'd like to get public meetings started in February for design overviews and bug scrubs. Stephano will be working with Ray to set these up. The hope is that we will have 1 meeting per month to start for bug scrubs. Design meetings will be dependent on how many design ideas have been submitted. The design meetings could also be used to discuss RFC's from the mailing list.
>>
>>
>> Thank you all for joining. As always, please feel free to email the list or contact me directly with any questions or comments.
>>
>> Kind Regards,
>> Stephano Cetola
>> TianoCore Community Manager
>>
>> _______________________________________________
>> edk2-devel mailing list
>> edk2-devel@lists.01.org
>> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.01.org%2Fmailman%2Flistinfo%2Fedk2-devel&amp;data=02%7C01%7Cjerecox%40microsoft.com%7Ce1986594f0094058f09208d68dcd9b0c%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636852311581795501&amp;sdata=%2ByLNjAyHNxw1oBxlH6wN%2BkWK38tP1OsD9n4kCzK1SVg%3D&amp;reserved=0
>> _______________________________________________
>> edk2-devel mailing list
>> edk2-devel@lists.01.org
>> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.01.org%2Fmailman%2Flistinfo%2Fedk2-devel&amp;data=02%7C01%7Cjerecox%40microsoft.com%7Ce1986594f0094058f09208d68dcd9b0c%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636852311581795501&amp;sdata=%2ByLNjAyHNxw1oBxlH6wN%2BkWK38tP1OsD9n4kCzK1SVg%3D&amp;reserved=0
> _______________________________________________
> edk2-devel mailing list
> edk2-devel@lists.01.org
> https://lists.01.org/mailman/listinfo/edk2-devel