Repository navigation
Conversation
…nd earlier Firedrake 2026 made Mesh.geometric_dimension an attribute; earlier releases expose it as a method. viscous_power and flow_law called it, so the inversion in icepack/ismip7 could not run on the 2026 environment without this edit, which has lived uncommitted in the editable installs at Rice and on one workstation since 9 June. Landing it so every site can clone rather than rsync (icepack/ismip7#46).
Member
Author
|
@dlilien @danshapero for review: four lines, the Firedrake 2026 geometric_dimension shim that icepack/ismip7#46 traced to two uncommitted editable installs. Tests: 13 passed on 2026.4.1. |
|
Good by me, I think a similar approach is in icepack1, but I'll leave this one to @danshapero to decide since there are other failing tests. |
Member
|
We can also change this to only read it as an attribute and change the test suite to run on firedrake-2026.4.2 or whatever the latest release is. We're going to have to drop support for old Firedrakes from mainline icepack soon anyway. |
dlilien
added a commit
to icepack/ismip7
that referenced
this pull request
Sep 22, 2026
The inversion needs two lines Firedrake 2026 forces on icepack2. They were uncommitted edits in one workstation checkout, rsynced to the clusters, so nobody could tell whether two sites ran the same code. They are now a commit on the branch of icepack/icepack2#3, which is icepack2 main (40e848b) plus those two files. install_deps.sh names that branch's head by SHA and clones it, and --check-icepack2 prints what a site has. A checkout that is not a clone, or that carries local changes, is reported and left alone: the rsynced copy may hold the only record of whatever else is in it. The three ICEPACK2_* variables follow the fix onto main when the pull request merges, and tests/test_icepack2_pin.py pins the defaults so that edit is deliberate. The two annotations pointing at issue #66 are retargeted now that it is closed: the re-measure the readiness rows owe belongs to issue #30. Closes #46 Co-Authored-By: Claude Opus 5 <[email protected]>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Firedrake 2026 made
Mesh.geometric_dimensionan attribute; earlier releases expose it as a method.viscous_powerandflow_lawcall it, so the ISMIP7 inversion could not run on the 2026 environment without this. The edit has lived uncommitted in the editable installs at Rice and on my workstation since 9 June (identical on both, md551db0898…), which is icepack/ismip7#46: two sites could not be shown to run the sameicepack2, and Quartz runs without it.Four lines, both call sites:
test/composite_rheology_test.pyandtest/momentum_balance_test.pypass on Firedrake 2026.4.1 (13 passed). Once this lands,install_deps.shcan clone rather than rsync and every site is on one committed state.