-
Notifications
You must be signed in to change notification settings - Fork 146
Minutes for 2021 meetings
Attendees: Mike Ek, Andy Hazelton, Dom Heinzeller, Jeff McQueen, Jeremy Gibbs, Jimy Dudhia, Ligia Bernardet, Lisa Bengtsson, Matt Dawson, Joe Olson.
Meeting notes now on the wiki page: https://github.com/NCAR/ccpp-physics/wiki/Minutes-for-2021-meetings. Slides: CCPP-Physics_Code_Mgmt_20211215: https://docs.google.com/presentation/d/16J_AzpL3CteNVgWKMEk7M-tm-YxBzEKcDF067xsmhOc/edit#slide=id.gea8fd115cd_0_17.
Update on POC for physics schemes.
Discuss CCPP visioning workshop: DTC proposal for AOP 2022 (slides 7-13 in Ligia's slides for today) Leverages NOAA-NCAR MOA. Goal is to be able to share the CCPP framework with NOAA and NCAR, also NRL (others that may adopt CCPP). Goals, workshop outcomes, audience, organizing committee, logistics, things to think about, and possible invited speakers. Likely not combined with a UFS physics workshop.
Comment from Andrew Hazelton - NOAA Affiliate2:27 PM For a DTC evaluation metrics workshop I attended, there were surveys sent out beforehand about key metrics. Maybe something similar could be created for key discussion topics for the CCPP and sent to all groups?
Per Andy's comment above, some "homework" beforehand so the we can come into the workshop with an efficient agenda/topics to cover. Familiarize the potential workshop attendees on what CCPP has to offer --for those now yet aware too much.
Lisa B.: different sessions depending on a given workshop attendee's experience with CCPP --from initial/beginner, to advanced.
Ligia and Mike working on a corresponding CCPP Visioning DTC proposal for AOP 2022.
Attendees: Ligia, Dom, Matt, Andrew, Jeremy, Lisa, Mike, Laurie, Jimy, Kate, Fanglin
Topic(s): Wiki page: will request OK from group to post meeting minutes (publicly available)
POC List is finalized, DTC will implement as GitHub code-owners
CCPP Requirements: Last modified in 2018, is out of date now; proposal to use CCPP Tech Doc and GitHub issues and projects to continue this effort, leave requirements doc as historical record. https://ccpp-techdoc.readthedocs.io/en/latest/ Action: review the current Tech Doc to see if updates or additions are needed in order to capture the requirements (for developers) Is the current Tech Doc “easy to find”? Links on the GitHub wiki (ccpp-physics) and in the DTC CCPP web/landing page
Code Management Recap: Plan would implement institutional forks for host models, not yet implemented EMC is hiring staff to assist with this role (managing EMC/UFS fork), also assist EMC developers with CCPP Should there be a single “UFS” fork, or actually one from each institution involved? (AOML, PSL, etc) I.e. EMC would not have an EMC fork, but could own/maintain the UFS fork for ufs-weather-model? Depends on what you’re working on, and who you’re working with Every fork needs to have a commitment to maintain that fork!
Next year: consider a CCPP vision workshop, to consider advanced options, how it should evolve. Fanglin: perhaps a joint workshop with UFS-physics? Matt: workshop also being planned to discuss aerosols/chemistry/host-model interfaces;
CHAT: You1:59 PM https://docs.google.com/presentation/d/11ZXDfqZjM36OrikAGYJ5aY_z5575Txz9DSZM66up6P8/edit#slide=id.p1 You2:02 PM https://docs.google.com/presentation/d/11ZXDfqZjM36OrikAGYJ5aY_z5575Txz9DSZM66up6P8/edit#slide=id.p1 https://docs.google.com/document/d/1Cq6OaejTF_DQcb0gTDIn7l707nyTp5w9RWoDtPBC3KI/edit Dominikus Heinzeller - NOAA Affiliate2:03 PM Hi Andy, welcome! Lisa Bengtsson - NOAA Affiliate2:05 PM Best of luck Dom and congratulations Laurie, you will both be missed! Ligia Bernardet - NOAA Federal2:16 PM https://ccpp-techdoc.readthedocs.io/en/latest/ Fanglin Yang - NOAA Federal2:20 PM Is it a good idea to provide a PDF version of the Technical Documentation for users to download? Andrew Hazelton - NOAA Affiliate2:22 PM That would be helpful (a PDF) Dominikus Heinzeller - NOAA Affiliate2:22 PM @Fanglin, on that website (readthedocs), you can click at the bottom right and download a PDF Fanglin Yang - NOAA Federal2:23 PM ok & thanks Dominikus Heinzeller - NOAA Affiliate2:31 PM @Fanglin Thank you for the credits and I totally agree with you. EMC needs a CCPP person. Lisa Bengtsson - NOAA Affiliate2:38 PM If we go in this direction, is it advised that PSL, GFDL, AOML, NSSL etc also create their institutional forks? Or should we work in main? or EMC or GSL repos? Andrew Hazelton - NOAA Affiliate2:39 PM With other applications (e.g. HAFS) we have been using the forks maintained by EMC Fanglin Yang - NOAA Federal2:50 PM In addition to GSL, who else has CCPP forks ? NRL ? NCAR ? Dominikus Heinzeller - NOAA Affiliate2:51 PM NRL has, yes - but we can't see their code. NCAR doesn't have its own fork, DTC had one in the past but this is no longer in use. Lisa Bengtsson - NOAA Affiliate3:01 PM Make sure Greg Frost is in the loop perhaps
Attendees: Ligia, Dom, Jim, Jimy, Matt, Jeff, Fanglin, Mike
Topic(s):
New members slide
Meeting Notes: Public? Consensus that it is OK. Ask broader group.
PoC Update: (Ligia) GFDL mp: Ruiyu, Linjiong is SME GFDL sfc layer: Zhan? Linjiong? Some with no response: Kruger, Yudin, EMC list for Fanglin to contact, also another email from Ligia to get written response
CCPP Requirements Doc: (Dom) Initial version Nov 2015 2018 added NCAR requirements Linked on wiki page Most requirements met by current CCPP, other new ones added, others not met. Needed updates listed. H6 not clear. Concurrently means all physics operate on same state or on parallel processors? Use of optional arguments, active flag, 3d physics in future Process for updating requirements Ligia: can be done by management or framework committees in consultation with each other. Discussed with Louisa - Committee not solely responsible, need to consult their own management for buy-in or for delegated authority to committee member as management is unlikely to be involved in technical details. Matt: Purpose of requirement doc after development is mostly done? Some are obsolete. Dom: maybe remove implemented requirements and leave future needs as a more agile document. Jimy: developers can just access a rules document and may not need full requirements Matt: updating needs can be done via a separate document. Ligia: could be more like an Issues list, more readable Fanglin: document not well known at EMC Consult Steve on why this document is needed and why it has to be maintained Documents: Requirements doc https://dtcenter.org/sites/default/files/community-code/ccpp-requirements-may2018.pdf Framework page with link to above https://dtcenter.org/community-code/common-community-physics-package-ccpp/requirements CCPP Tech Doc with Rules for development: https://ccpp-techdoc.readthedocs.io/en/latest/CompliantPhysicsParams.html
-end
physics points-of-contact https://github.com/NCAR/ccpp-physics/wiki/CCPP-Physics-Code-Management List of participating organizations (not individuals), representatives from each are in this call (invited, may not attend each time!) No further comments from the group
https://github.com/NCAR/ccpp-physics/wiki/CCPP-Physics-Points-of-Contact Is there a veto option? If a difference of opinion occurs, who ultimately decides whether a change goes in or not Generally, this will be resolved with consensus, but in some cases, there might be unique forks with a very specific code modification Will not try to specify bureaucratic rules at this level Solution may be to add a switch/option Should the rules also include the “reject” option (re: POCs either “approve” or “request changes”…) Wording added to the wiki page
Draft email to nominees (Ligia emailed earlier today, Wed 3 Nov 2021 8:32am MT) This might be a surprise for some people, may need some iteration with supervisors, etc Reviewing/refining Ligia’s email: Message to be sent to CCPP Physics POCs (column E) in “Draft CCPP Primary Schemes Points of Contact”: https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0. Latest draft of this email: From: Ligia Bernardet To: All PoCs (blind copy) Subject: Invitation to serve as CCPP Physics primary scheme(s) Point of Contact Dear CCPP contributor, On behalf of the CCPP Physics Code Management Committee, which has representatives from the major organizations contributing to CCPP, I would like to invite you to serve as Point of Contact (PoC) for one or more primary schemes in the CCPP physics authoritative repository on GitHub (https://github.com/NCAR/ccpp-physics). We are inviting you to serve in this capacity because you have been identified as a significant stakeholder for the scheme(s). Please search for your name on column E in the list of nominated PoCs to find which scheme(s) we are asking you to represent. Note that column F lists primary developer(s) and/or subject-matter expert(s) for the schemes. You are welcome to add additional names to column F. Scheme PoC is a new role we are establishing to foster community in the development of primary physical parameterizations by creating a friendly environment for collaborative development, create long-term vision for the distribution and support of primary physical parameterizations, and prevent uncoordinated changes to schemes. A description of what this role entails is available in the PoC section of the CCPP Physics code Management wiki page. If you have any questions about this request or would like to get more information and discuss it, please reach out to me. We hope that you can let us know no later than November 12th, 2021. If you cannot serve in this role, I would appreciate it if you could suggest an alternate. Best regards, Ligia Bernardet
CHAT: Ligia Bernardet - NOAA Federal2:02 PM https://github.com/NCAR/ccpp-physics/wiki Mike Ek2:03 PM Draft CCPP Primary Schemes Points of Contact: https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0 Ligia Bernardet - NOAA Federal2:12 PM We are discussing this wiki page: https://github.com/NCAR/ccpp-physics/wiki/CCPP-Physics-Points-of-Contact David Gill2:29 PM SME? Ah Dominikus Heinzeller - NOAA Affiliate2:30 PM Yes, better David Gill2:59 PM Dropping off, nice job Ligia
Attendees: Ligia Bernardet, Mike Ek (notes), Dom Heinzeller, Jim Doyle, Jimmy Dudhia, Joe Olson, Lisa Bengtsson, Louisa Nance, Matt Dawson, Fanglin Yang
Topic(s): Spreadsheet with PoCs (“Draft CCPP Primary Schemes Codeowners”): https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0 Ligia’s slides: https://meet.google.com/tsj-rscn-krm?authuser=0)
Notes: (as we go through Ligia’s slides) Revisit CCPP physics POCs Provided an overview of the motivation for the process for this group, and the method to make it work (more than just PRs!) Re-defined the PoCs primary roles/responsibilities --”steward” of a given scheme Practicalities for Primary Responsibilities: no change from last time this slide was shown. PoCs add’l roles/responsibilities: support CCPP team with planning for maintenance, release and support of a given scheme. Perhaps once or twice a year engagement on a given scheme. Review of the spreadsheet Comments: Lisa --helps provide “structure” for the CCPP physics schemes Jimy-- Morrison-Gettelman MP scheme: POC Anning Cheng (EMC)--more of a UFS focus, while the original developers: Hugh Morrison and Andrew Gettelman (NCAR) --not UFS focused. Matt Dawson (chemistry, NCAR/ACOM) --reasonable set of asks Joe Olson (PBL, GSL) --also a reasonable set of asks
Reviewing the spreadsheet and POCs: A number of CCPP physics schemes in the spreadsheet… some being considered for operations, many others still in research versions. No more than two POCs” per scheme. (otherwise perhaps too burdensome) In the spreadsheet, role of POCs more formalized, “Additional experts, contributors, original developer” --someone for the POCs to reach out to with questions/for add’l input. From the slide deck today (CCPP-Physics_Code_Mgmt_20211020), create a document about the PoCs roles/responsibilities --to reside in GitHub wiki Next mtg (Wed 03 Nov 2021), finish the review of PoCs/add’l names (we left off at row 25/26 --NRL related)
CHAT: Mike Ek 2:02 PM https://docs.google.com/document/d/1Cq6OaejTF_DQcb0gTDIn7l707nyTp5w9RWoDtPBC3KI/edit Ligia Bernardet - NOAA Federal2:02 PM Slides for today: https://docs.google.com/presentation/d/18Ykxm3LtUG1x135KsQUVoXu6gKukkRO0cCCPfXSakAM/edit#slide=id.p1 Mike Ek 2:10 PM Spreadsheet with nominees: https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0 Ligia Bernardet - NOAA Federal2:18 PM Spreadsheet of nominees: https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0 Ligia Bernardet - NOAA Federal2:20 PM Slides for today: https://docs.google.com/presentation/d/18Ykxm3LtUG1x135KsQUVoXu6gKukkRO0cCCPfXSakAM/edit#slide=id.gf5f7c22109_0_7 Lisa Bengtsson - NOAA Affiliate2:29 PM Don Dazlich colorado state univ works with Dave Randall Dominikus Heinzeller - NOAA Affiliate2:31 PM Agreed, no more than two POCs Jimy Dudhia2:31 PM github lists contributors which can give clues for who is active with a scheme James Doyle2:51 PM Sorry - have to drop off early for another call. Lisa Bengtsson - NOAA Affiliate2:53 PM For Thomson MP perhaps add Trude Eidhammer in the "contributor/orig develop" column? Louisa Nance3:00 PM makes sense - not everyone has access to google
Attendees: Ligia, Laurie, Mike, Jim, Dom, Lisa, Jeremy, Jimmy, Louisa, Joe, Jeff, Fanglin
Topic(s): CCPP Primary scheme code owners Slides from Ligia: https://docs.google.com/presentation/d/1eWs3NkkBVyMLfg22YSvEt6CGaY0kUi5JSbVrPhxx4yk/edit#slide=id.p1 Spreadsheet with nominees: https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0
Code owners responsibility:
Add an expected level of effort and/or priority, to make the role more palatable? It’s a long list!
Open question on timeliness - could it be more scheduled (quarterly, monthly, etc?) doesn’t work if it’s needed for pre-implementation, for example
Will this vary if the development first goes to an institutional fork? Where does the review happen, then, i.e. the GSL fork, or the ccpp-physics authoritative repo?
Motivation slide seems operational-center specific, how to make it more open to community developers, foster a friendly environment for contributions
Need to ensure code integrity - this was discussed earlier in code management context (reg tests, etc), not directly related to code owners.
Institutions may/will use a fork to manage their own contributions, code management is essential to maintain forks
Point-of-Contact role is accepted as part of the review process for the authoritative repository for ccpp-physics
Follow-up this discussion with a broader announcement to ensure all voices are heard, response by the next meeting
Code owners proposed list: Discussions around specific PoCs, what is a reasonable expectation for some of the original developers (likely to review/respond, or not?) Add a column for “original developer” or “additional expert”? Ask for all representatives to review this spreadsheet (again!)
CHAT: Michael Ek2:00 PM CCPP-physics-code-mgmt-mtg-notes: https://docs.google.com/document/d/1Cq6OaejTF_DQcb0gTDIn7l707nyTp5w9RWoDtPBC3KI/edit Ligia Bernardet - NOAA Federal2:00 PM https://docs.google.com/presentation/d/1eWs3NkkBVyMLfg22YSvEt6CGaY0kUi5JSbVrPhxx4yk/edit#slide=id.p1 Ligia Bernardet - NOAA Federal2:03 PM Slides: https://docs.google.com/presentation/d/1eWs3NkkBVyMLfg22YSvEt6CGaY0kUi5JSbVrPhxx4yk/edit#slide=id.p1 Michael Ek2:03 PM CCPP-physics-code-mgmt-mtg-notes: https://docs.google.com/document/d/1Cq6OaejTF_DQcb0gTDIn7l707nyTp5w9RWoDtPBC3KI/edit Lisa Bengtsson - NOAA Affiliate2:40 PM @Dom, does GSL fork all submodules in ufs-weather-model, or only ccpp-physics? Jimy Dudhia2:45 PM The slides could serve as minutes for others James Doyle2:47 PM Sounds like a good plan to have an email summary Ligia Bernardet - NOAA Federal2:49 PM https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0 Louisa Nance2:57 PM Recommend that you make an action item for next meeting for all reps to review this list and provide feedback on the proposed POCs Jimy Dudhia2:59 PM additional experts would be a good column Lisa Bengtsson - NOAA Affiliate2:59 PM yes, I agree with that Jimy James Doyle3:00 PM Have to drop off for another meeting Ligia Bernardet - NOAA Federal3:01 PM THANK OU!! YOU
Attendees: Ligia, Mike, Dom, Steve, Laurie, Jim, Jeff, Jeremy, Jimy, Joe, Lisa, Matt, Louisa, Kate, Fanglin
Topic(s): CCPP Primary scheme code owners Slides from Ligia: https://docs.google.com/presentation/d/1Lxb_i3kZ-W2uqGal1xvDmsBHOLpXV4lmr0vc6n7ONVw/edit#slide=id.p1
Code owners responsibility:
add a requirement for documentation, esp related to code best practices and implementation (as well as science doc)
Provide input for release planning (readiness)
Should there also be an overall “physics” code owner for code reviews, etc, for consistency - since ccpp-physics is its own repo, this is effectively in place as the repo owners
Code owners proposed list: https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0 Some concerns about who is currently assigned as code-owners, esp, EMC staff (FangLin) GitHub uses the Code Owners terminology to represent who is responsible for code reviews and PR approvals Responsibilities beyond this - can range from original developers to current maintainers Stochastic physics: this “scheme” also uses code outside of CCPP; interacts with other schemes, perhaps like an “HWRF” flag? An outlier (esp. C-A) How to handle development outside of the GFS/EMC repo, that still needs to coordinate with the authoritative/operational codes - where would code owners be involved in this process? (PUMA, NOAH-MP, etc) Each repository has to have its own governance process, how they interact is also important Should we (will we) also have a committee to review physics, not “just” the code owner for each scheme; decides retirement, addition, etc There may need to be some education, GitHub procedures, etc Consider the time commitment (and $$) for this role - should there be a consistent approach across institutions?
NEXT STEPS:
Add comments and/or update names, 2 weeks then finalize
Also add any details about major-namelist-switches, secondary Fortran files as part of a scheme.
Implement this in the ccpp-physics repository
CHAT: Ligia Bernardet - NOAA Federal2:03 PM https://docs.google.com/presentation/d/1Lxb_i3kZ-W2uqGal1xvDmsBHOLpXV4lmr0vc6n7ONVw/edit#slide=id.p1
Mike Ek2:08 PM Laurie: since we have Ligia's slides, I suppose note taking is any questions that come up during the presentation, in the chat, and any ensuing discussion afterwards.
Laurie Carson2:17 PM https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0
Ligia Bernardet - NOAA Federal2:25 PM https://docs.google.com/presentation/d/1Lxb_i3kZ-W2uqGal1xvDmsBHOLpXV4lmr0vc6n7ONVw/edit#slide=id.gea8fd115cd_0_17
Jimy Dudhia2:28 PM code owner is github terminology
Ligia Bernardet - NOAA Federal2:30 PM Spreadsheet: https://docs.google.com/spreadsheets/d/14y0Th_sSpCqlssEMNfSZ_Ni9wrpPqfpPY0kRG7jCZB8/edit#gid=0
Fanglin Yang - NOAA Federal2:54 PM Jim, good point. [Jim: Code developers/owners may not be paid to engage in CCPP-related activities, although ultimately it would be in their best interest.]
Attendees: Ligia, Mike, Dave, Joe, Matt, Jeremy, Louisa, Jimy, Kate, Jim, Steve, Dom, Laurie, Jeff, Fanglin, and Laura Fowler.
Topic(s): Requirements (and coding) to ensure that a given set of MYNN PBL low level routines can run in both the UFS and MPAS. This pioneer work opens the door to practical CCPP interoperability between modeling systems and makes our discussions about a proposed shared repository for physics to be used by various models more palpable and less abstract.
Notes
Joe presented a Google Doc with complexities and challenges of sharing a MYNN scheme between various host models using CCPP. 1) The scheme will require either an interface or a set of interstitials to work with diff dycores, which have diff state variables (e.g., spec hum vs mix rat). Goal is to have the “scheme itself” to be the same in all dycores. 2) Need dependencies/consistency checks (e.g., WRF check_a_mundo). 3) Need a mechanism to know which microphysics species are coming in the PBL (CCPP MYNN has interstitial, WRF has pbl_scalar_check). 4) Configurable TKE options, esp. use or not of advected TKE, and whether or not to allocate some arrays. 5) Fractional surface types. 6) In different models, the schemes perform different duties (e.g., sfc layer performs or not diagnostic of 2-m T). Diagnostic output: should it be packaged and labelled individually (more transparent to users) or assembled in a single 4D-array or DDT (handy bundle for developers)? For Thompson, we bundled the diagnostic arrays - this is less transparent (as there is not metadata for each variable) but that does not matter because the diagnostic arrays are just for the scheme developers, who know what they are. Fangling said that Ligia/Dom’s letter-of-intent to JTTI funding about “surface composites” may help make this problem more tractable. Status: After Ligia/all clarified concepts of primary vs interstitial vs wrappers in a previous meeting, Joe/Laura went back to work. Most challenges with MYNN have already been addressed. The next-gen CCPP Framework code generator (capgen.py) differs from current code ccpp_prebuild.py wrt handling allocation of physics arrays. Currently, the host has to do all the allocation of arrays. With capgen, physics will “control” allocation. Dave Gill: There are advantages for code developers to have their schemes be CCPP-compliant so they do not need to maintain multiple versions (in CCPP, WRF, MPAS etc.). This will reduce versionites. And everybody will know where to get the latest version. We (Laura) now have a generalized MYNN scheme that works without the CCPP-Framework it - does not have metadata yet. Metadata info is of big assistance to users/developers. Discipline myopia: a scheme that was used for a given application (say, super high-res) will face challenges when applied in a different situation. How do we decide which schemes get accepted onto CCPP-Physics - need an engaged author. Get requirements. Different missions of op centers, main authors, and other developers. Changes will improve here and degrade there - so how do we decide? How to handle repository(ies) - explosion of repos is problematic. I (Dave) am not supportive of having each scheme in its own repo (a possi9. What is an atmos physics scheme (we have ocean/ice/etc.)? Consistency checks should be outside of CCPP. Laura: Work has been on creating a clean interstitial layer - decided against a wrapper. Part of the work was initialization of constants. Clean version of PBL scheme is now in use. Sfc layer still has a challenge related to sfc type decomposition. Joe: Different dycores will need different interstitials, which means different dycores need different SDFs. So, SDFs mill multiply. Also SDFs multiply with suite variations (that is, using MYNN PBL with other schemes). Dave: It is the responsibility of a host model person to bring in a CCPP-compliant scheme. To use a CCPP-compliant scheme in a given host model, host models will have to provide their own interstitials. Next steps, actions. Create guidelines for bringing a scheme into CCPP. Joe: should it include how to code the host model? That is, what to do in GFS_typedefs? Jim: NRL may have feedback on this process Dave: CCPP is not a turn-key solution. Dom: host model people and developers will always have to work together to solve problems. Will always have to be a collaborative effort.
Chat capture
Michael Ek2:04 PM CCPP-physics-code-mgmt-mtg-notes: https://docs.google.com/document/d/1Cq6OaejTF_DQcb0gTDIn7l707nyTp5w9RWoDtPBC3KI/edit Steve Goldhaber2:32 PM Sorry, I'm late for another meeting. You (Ligia) 2:34 PM @Joe can your Google Doc be shared with meeting participants? You2:36 PM @Joe I can also share it if it is easier for you. Michael Ek2:41 PM @Joe @Ligia Will you share the link to the google doc here? Joe's google doc Joseph Olson - NOAA Federal2:42 PM https://docs.google.com/document/d/1uMQ3hMhKD_dNGhkZiryhgcJwBRjIjFZmft7adpjVbZM/edit?usp=sharing Michael Ek2:44 PM @Joe. I also saw that you shared via email. Thanks. James Doyle2:58 PM NRL might have some comments too to add to the doc based on our experience with NEPTUNE and implementing CCPP physics. Send a message to everyone send Send message CCPP physics management bi-weekly meeting
Attendees: Ligia, Mike, Dom, Laurie, Jimy, Lisa, Matt, Louisa, Jeremy, Joe Olson, Jim Doyle, Jeff McQueen, Fang Lin
Topic: Code ownership for schemes in ccpp-physics repo
Ligia’s slides
Discussion: May need multiple code owners, include the original developer and a representative of the operational centers. WRF (and MPAS): most schemes have an “expert” in mind, usually the original developer, informal (not GitHub code owner mechanism) Case: original author basically abandoned their “scheme”, so no one was an obvious code owner anymore Jim Doyle: idea is good, but practicality could be problematic - what is the response from the original developers? Jimy (WRF) - been fortunate so far, most will accept bug fixes quickly, haven’t had trouble with owners not wanting to accept updates (so far) Fang Lin: “ownership” cannot include copyright for NOAA staff - which it doesn’t, just a review/approval step; operational centers have a unique role in this question Lisa: have there been cases of code divergence when the original author didn’t want to include new proposed development? Not so far, most of these are bug fixes (Jimy) Mike: should focus on collaborative roles, even though there may be occasional conflicts Jim: Navy also has operational concerns, similar to EMC; this may seem like adding layers to the process, but should also recognize that this is enabling a welcome collaboration that wouldn’t happen otherwise Joe: interactions with community contributions: a wide variety of quality and applicability, some are well-coded, well thought out, others not so much; it’s time consuming to review these contributions (incl testing) Fang Lin: another potential issue is changes proposed that may work well for global but not regional, who decides?
Next Steps: Propose a list of code owners for each primary schemes for further discussion here (DTC)