Adding psi4 and dkh #22328

loriab · 2023-03-18T23:27:03Z

Psi4 is a quantum chemistry package in Python and C++ with pybind11 bindings. It's been conda-packaged for a long time on its own channel (https://anaconda.org/psi4/psi4; recipe https://github.com/psi4/psi4meta/tree/master/conda-recipes/psi4-multiout), so this is finally a port to conda-forge, now that all required dependencies are available. The whole optional dependency ecosystem isn't ported yet, but I'm testing one of every language (C/Py: pylibefp; Fortran: dkh (alongside)), so I think this recipe is stable and ready for review.

The primary concern I anticipate is that this isn't packaging an upstream release, only an upstream branch. This is because the psi4 conda package and developer build instructions have always been based on -c defaults -c psi4, but there's one big and accumulating little changes to prepare for c-f deps, and I'd like to preserve the usability of the existing instructions until the whole software stack is ready on c-f. So if this recipe is merged, I expect that the first release from a c-f feedstock will have an ugly version (e.g., 1.8a2). I'd be glad to have it on a dev branch and psi4_dev channel if that can be done at feedstock-creation-time. A c-f compatible psi4 v1.8 release is due for late April/early May.

Checklist

Title of this PR is meaningful: e.g. "Adding my_nifty_package", not "updated meta.yaml".
License file is packaged (see here for an example).
~~Source is from official source.~~ (See notes above)
Package does not vendor other packages. (If a package uses the source of another package, they should be separate packages or the licenses of all packages need to be packaged).
~~If static libraries are linked in, the license of the static library is packaged.~~ (No static libs)
Package does not ship static libraries. If static libraries are needed, follow CFEP-18.
Build number is 0.
A tarball (url) rather than a repo (e.g. git_url) is used in your recipe (see here for more details).
GitHub users listed in the maintainer section have posted a comment confirming they are willing to be listed there.
When in trouble, please check our knowledge base documentation before pinging a team.

conda-forge-webservices · 2023-03-18T23:27:22Z

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipes/dkh) and found some lint.

Here's what I've got...

For recipes/dkh:

There are 1 too many lines. There should be one empty line at the end of the file.

conda-forge-webservices · 2023-03-19T03:17:05Z

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipes/dkh) and found it was in an excellent condition.

conda-forge-webservices · 2023-03-20T17:28:31Z

Hi! This is the friendly automated conda-forge-linting service.

I wanted to let you know that I linted all conda-recipes in your PR (recipes/psi4) and found some lint.

Here's what I've got...

For recipes/psi4:

Failed to even lint the recipe, probably because of a conda-smithy bug 😢. This likely indicates a problem in your meta.yaml, though. To get a traceback to help figure out what's going on, install conda-smithy and run conda smithy recipe-lint . from the recipe directory.

conda-forge-webservices · 2023-03-20T18:12:19Z

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipes/psi4) and found it was in an excellent condition.

loriab · 2023-03-28T18:16:52Z

@conda-forge/help-c-cpp, ready for review!

fevangelista · 2023-04-04T14:13:28Z

I am happy to help maintain this recipe.

recipes/psi4/conda_build_config.yaml

recipes/psi4/meta.yaml

carterbox · 2023-04-18T18:50:31Z

recipes/psi4/meta.yaml

+    - {{ pin_compatible('optking', upper_bound='0.5') }}
+    - {{ pin_compatible('qcelemental', upper_bound='0.50') }}
+    - {{ pin_compatible('qcengine', upper_bound='0.50') }}


There are static linking / overdepending warnings in the logs for these libraries. Can you explain why they are needed at both runtime and build time if they aren't dynamically linked?

https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=690074&view=logs&j=6f142865-96c3-535c-b7ea-873d86b887bd&t=22b0682d-ab9e-55d7-9c79-49f3c3ba4823&l=13902

These three are pure-python packages with no run_exports. Their important role is in requirements/run, and the upper bound is because there's some API-breaking changes planned in the next year, and I wanted to isolate the built packages from those. Because these three are required dependencies, if they're not present at cmake configure-time, psi4 will download and vendor them, so their presence in requirements/host avoids that.

I think you're going to run into unresolvable dependencies in the future if you do it this way because eventually the host environment will move to optking =0.6.0 and then the run environment pinning expression will become optking >=0.6.0,<=0.5?

I think you want something like this:

host: - optking <0.6 run: - {{ pin_compatible('optking', max_pin='x.x') }}

Such that opking is always <0.6 and the same minor version in the host and run environments.

Thanks for the analysis. Agreed that the original setup will take a little management of host v run to avoid the situation you describe. optking is fairly new (Dec 2022) and has a much faster release cycle (whenever) than psi4 (twice yearly), so I wanted to leave room for optking bugfixes that psi4 could access (so not pinning to x.x) and also put an upper bound so if optking finds it needs to do a breaking change, this version of psi can be insulated. It's some management, but I'm pretty sure an incompatibility will show up at build-time for me, not at solve-time for user, so I'm ok with it if you are.

I moved the upper bound for optking out to v0.15, so all three have 12+ versions of space before hitting the limit.

I meant this comment to apply to all three of optking, qcelemental, and qcengine. Manually specifying upper_bound is just a very unusual use of these pinning expressions.

I wanted to leave room for optking bugfixes that psi4 could access (so not pinning to x.x)

Pinning to x.x does allow for bugfixes? For example:

host: - optking 0.5.2 run: - {{ pin_compatible('optking', max_pin='x.x') }} # This template would be replaced with optking >=0.5.2,<0.6.0

Are the bugfixes for optking not shipped via increment of the patch (major.minor.patch) version number? For projects using SEMVER, breaking changes should never happen at the patch level.

EDIT: Hmmm... I linked the semver documentation, but it seems that according to the spec with major version 0, anything breaking can happen at any time. I typically still never break anything when bumping the patch version.

To allow for the widest compatability (accept bugfixes) whilst protecting against API breaks, you should use max_pin='x.x' for a package whose major version is zero. You can always rebuilt psi4 when new versions of these packages become available.

I should have said "room for bugfixes and development". optking is new-ish and likely to be fast moving, so I was just being cautious and leaving room for an API breaking change less momentous than 1.0. I can release the constraint if you like.

The qcelemental and qcengine are in a stack with qcfractal (presently at v0.15.8 on c-f). But the last has an imminent API breaking change at 0.50 (currently in development off-c-f at https://anaconda.org/qcarchive/qcfractal/files), and once that's ready, qcelemental and qcengine will have a similar scheduled API break that we'll align at v0.50 for consistency. So I think those two are important to keep.

fwiw, I've used the upper_bound pattern before, and it's worked cleanly for the users (all their complaints come to me :-) .

carterbox · 2023-04-18T18:51:59Z

recipes/dkh/meta.yaml

@@ -0,0 +1,71 @@
+{% set name = "dkh" %}


DKH looks good in both recipe and logs; good job!

Co-authored-by: Daniel Ching <carterbox@users.noreply.github.com>

loriab · 2023-04-19T03:35:37Z

Thanks very much for the review and improvements, @carterbox. I think I've addressed the comments, and I went ahead and added a third-party-licences file here, too, for a few miscellaneous files. CI will be running for a couple hours, but a one-python-version build was clean.

loriab · 2023-04-25T23:26:45Z

Sorry to ping you @carterbox, but is there anything I can do to facilitate this PR? I wouldn't usually push, but I've got a release date for psi4 looming. Thanks for all your work reviewing these.

carterbox

I'm adjusting your pinning templates to be more conservative, so I can merge this PR today. These pinnings will better protect against API breaks. Just rebuild the package whenever new versions of any these dependencies are released.

recipes/psi4/meta.yaml

loriab · 2023-04-26T03:14:25Z

I'm adjusting your pinning templates to be more conservative, so I can merge this PR today. These pinnings will better protect against API breaks. Just rebuild the package whenever new versions of any these dependencies are released.

Sounds good, thanks for the analysis, caution, and approval. :-)

dkh recipe

0282c93

loriab added 3 commits March 18, 2023 22:05

iter 2

526f444

iter 3

c567a38

iter 4

408fdce

initialize psi4 recipe

5ecbba6

loriab added 2 commits March 20, 2023 13:53

take 2

2380ca6

take 3

5bcdfbb

loriab added 19 commits March 20, 2023 14:24

take 4

4d0625b

take 5

0d9af60

take 6

978610d

take 7

fb74ed0

take 8

e87b657

take 9

c30fcf5

take 10

cdc9bd0

take 11

b7c7164

take 12

599318a

take 13

f7a1aa9

take 14

f91e70f

take 15

a7168c5

take 16

6f2d136

take 17

0599204

take 18

54a2e09

take 19

8c3695f

take 20

790672c

take 21

a796b80

take 22

d0854fb

iter 72

584072c

github-actions bot added c-cpp review-requested labels Mar 28, 2023

loriab mentioned this pull request Mar 28, 2023

capabilities overlap lilyminium/psiresp#104

Closed

loriab added 3 commits April 4, 2023 20:23

iter 73

5c60ec2

iter 74

614c97a

iter 75

673331b

loriab mentioned this pull request Apr 6, 2023

Add PCMSolver #22286

Merged

13 tasks

isuruf reviewed Apr 14, 2023

View reviewed changes

recipes/psi4/conda_build_config.yaml Outdated Show resolved Hide resolved

iter 76 -- review

88c042a

carterbox requested changes Apr 18, 2023

View reviewed changes

carterbox removed the review-requested label Apr 18, 2023

github-actions bot added the Awaiting author contribution label Apr 18, 2023

loriab and others added 4 commits April 18, 2023 19:07

Update recipes/psi4/meta.yaml

5e6ee18

Co-authored-by: Daniel Ching <carterbox@users.noreply.github.com>

Update recipes/psi4/meta.yaml

9a528d5

Co-authored-by: Daniel Ching <carterbox@users.noreply.github.com>

iter 77 -- review

a2440bb

iter 78 -- review

0298779

loriab requested a review from carterbox April 20, 2023 13:54

loriab added review-requested and removed Awaiting author contribution labels Apr 20, 2023

carterbox approved these changes Apr 26, 2023

View reviewed changes

carterbox reviewed Apr 26, 2023

View reviewed changes

recipes/psi4/meta.yaml Outdated Show resolved Hide resolved

Use conservative pinnings

2fb6419

carterbox merged commit b366f5d into conda-forge:main Apr 26, 2023

loriab deleted the loriab-psi4 branch April 26, 2023 03:15

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Adding psi4 and dkh #22328

Adding psi4 and dkh #22328

loriab commented Mar 18, 2023 •

edited

Loading

conda-forge-webservices bot commented Mar 18, 2023

conda-forge-webservices bot commented Mar 19, 2023

conda-forge-webservices bot commented Mar 20, 2023

conda-forge-webservices bot commented Mar 20, 2023

loriab commented Mar 28, 2023

fevangelista commented Apr 4, 2023 •

edited

Loading

carterbox Apr 18, 2023

loriab Apr 18, 2023

carterbox Apr 18, 2023 •

edited

Loading

loriab Apr 19, 2023

loriab Apr 19, 2023

carterbox Apr 26, 2023 •

edited

Loading

carterbox Apr 26, 2023

loriab Apr 26, 2023

carterbox Apr 18, 2023

loriab commented Apr 19, 2023

loriab commented Apr 25, 2023

carterbox left a comment

loriab commented Apr 26, 2023

Adding psi4 and dkh #22328

Adding psi4 and dkh #22328

Conversation

loriab commented Mar 18, 2023 • edited Loading

conda-forge-webservices bot commented Mar 18, 2023

conda-forge-webservices bot commented Mar 19, 2023

conda-forge-webservices bot commented Mar 20, 2023

conda-forge-webservices bot commented Mar 20, 2023

loriab commented Mar 28, 2023

fevangelista commented Apr 4, 2023 • edited Loading

carterbox Apr 18, 2023

Choose a reason for hiding this comment

loriab Apr 18, 2023

Choose a reason for hiding this comment

carterbox Apr 18, 2023 • edited Loading

Choose a reason for hiding this comment

loriab Apr 19, 2023

Choose a reason for hiding this comment

loriab Apr 19, 2023

Choose a reason for hiding this comment

carterbox Apr 26, 2023 • edited Loading

Choose a reason for hiding this comment

carterbox Apr 26, 2023

Choose a reason for hiding this comment

loriab Apr 26, 2023

Choose a reason for hiding this comment

carterbox Apr 18, 2023

Choose a reason for hiding this comment

loriab commented Apr 19, 2023

loriab commented Apr 25, 2023

carterbox left a comment

Choose a reason for hiding this comment

loriab commented Apr 26, 2023

loriab commented Mar 18, 2023 •

edited

Loading

fevangelista commented Apr 4, 2023 •

edited

Loading

carterbox Apr 18, 2023 •

edited

Loading

carterbox Apr 26, 2023 •

edited

Loading