rxd handles np.float64 - #3808
Merged
Merged
Conversation
|
✔️ 31e8684 -> Azure artifacts URL |
adamjhn
approved these changes
Jun 25, 2026
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## master #3808 +/- ##
==========================================
- Coverage 68.44% 68.43% -0.02%
==========================================
Files 688 688
Lines 111309 111311 +2
==========================================
- Hits 76190 76175 -15
- Misses 35119 35136 +17 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
ramcdougal
enabled auto-merge (squash)
June 25, 2026 01:53
|
✔️ 95affc9 -> Azure artifacts URL |
|
nrnhines
added a commit
that referenced
this pull request
Jul 21, 2026
* Fix Python extension library name (#3679) When generating the name of the NEURON Python libraries (like `hoc.cp313.so`), we make use of `CMAKE_LIBRARY_ARCHITECTURE`. On Fedora with the clang compiler, `CMAKE_LIBRARY_ARCHITECTURE` is set to `x86_64-redhat-linux-gnu`, but Python does not recognize this extension when importing: ``` $ python -c 'import importlib.machinery;print(importlib.machinery.EXTENSION_SUFFIXES)' ['.cpython-313-x86_64-linux-gnu.so', '.abi3.so', '.so'] ``` The solution is to not rely on `CMAKE_LIBRARY_ARCHITECTURE` and just always set the extension to `${ARCH}-linux-gnu`, which should be importable in Python on all "reasonable" Linux platforms. * Bump urllib3 from 2.5.0 to 2.6.0 in /ci (#3681) * Bump urllib3 from 2.5.0 to 2.6.0 in /ci Bumps [urllib3](https://github.com/urllib3/urllib3) from 2.5.0 to 2.6.0. - [Release notes](https://github.com/urllib3/urllib3/releases) - [Changelog](https://github.com/urllib3/urllib3/blob/main/CHANGES.rst) - [Commits](urllib3/urllib3@2.5.0...2.6.0) --- updated-dependencies: - dependency-name: urllib3 dependency-version: 2.6.0 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> * Ownership --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Goran Jelic-Cizmek <goran.jelic-cizmek@epfl.ch> * Add NMODL development guide (#3683) * Bump urllib3 from 2.6.0 to 2.6.3 in /ci (#3693) * Bump urllib3 from 2.6.0 to 2.6.3 in /ci Bumps [urllib3](https://github.com/urllib3/urllib3) from 2.6.0 to 2.6.3. - [Release notes](https://github.com/urllib3/urllib3/releases) - [Changelog](https://github.com/urllib3/urllib3/blob/main/CHANGES.rst) - [Commits](urllib3/urllib3@2.6.0...2.6.3) --- updated-dependencies: - dependency-name: urllib3 dependency-version: 2.6.3 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> * Retrigger CI --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: nrnhines <michael.hines@yale.edu> * NMODL: Handle `state_discontinuity` call in NET_RECEIVE block (#3592) * Fix the Neumann BC in the x-direction of fixed step extracellular diffusion. (#3690) * Update ECS node volume access and testdata. * Update data for neuron.test_rxd(). * Bump bokeh from 3.7.2 to 3.8.2 in /ci (#3691) Bumps [bokeh](https://github.com/bokeh/bokeh) from 3.7.2 to 3.8.2. - [Changelog](https://github.com/bokeh/bokeh/blob/3.8.2/docs/CHANGELOG) - [Commits](bokeh/bokeh@3.7.2...3.8.2) --- updated-dependencies: - dependency-name: bokeh dependency-version: 3.8.2 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: nrnhines <michael.hines@yale.edu> * NMODL: Add codegen for implicit `diam` and `area` (#3631) * Add explicit METHOD to SOLVE if target is a DERIVATIVE block (#3600) * Add unrolling of FROM iteration variables (#3597) * NMODL: Generalize MergeInitialBlocksVisitor (#3598) * Add tests * Fix name collision when merging top-level blocks (#3701) --------- Co-authored-by: David McDougall <dam1784@rit.edu> * link updates to readthedocs (#3718) * How to contribute to Documentation (#3721) * Update forum link to NEURON discussion board (#3724) * Bump tornado from 6.5.1 to 6.5.5 in /ci (#3723) * ci: add conditional Codecov token for Dependabot branches --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: nrnhines <michael.hines@yale.edu> * link back to forum (#3726) * ci: for macosx wheels, try to use system clang. (#3730) * This allows minimization of MACOSX_DEPLOYMENT_TARGET Otherwise linkage to libunwind with brew clang requires a deployment target of 14 * wheel build on macos-15 runners: ensure libomp installed * Path has specifc paths for brew bison and flex * fix: make Sprintf interpret %% -> % correctly when no args are passed (#3727) * FAQ comment about GUI. (#3738) * NMODL: make v, celsius, and dt available in VERBATIM (#3740) * more neutral language (#3739) * Bump minimal python to 3.10 (#3745) * 1.9 -> 1e9 for large dt (#3754) * removed link to page that doesn't exist in current docs (#3731) * CTNG cleanup (#3564) * refactoring into smaller functions and 2->3 updates * removed legacy code * removed voxelize and voxelize2 from cmakelists * corner_pts a list, removed circular imports * cython shortcuts and minor cleanups * don't use bisect * removed numpy array creation from closest_pt and extreme_pts * don't need numpy for isnan * xs,ys,zs are numpy array; using memview; removed unused function * changed % formatting to f-string * updated cmake documentation link (#3775) * Python typo fix (#3776) comma was period * fix section pushing examples in C API docs (#3785) * handle segment iteration when no mechs (#3794) * Fix Vector.min_ind(i1, i2) bug (#3787) * test of Vector.min_ind(i1, i2) * Doc warning for Vector.min_ind with example * SourceForge link is unreliable in CI due to Cloudflare protection * fix for file type detection on linux and mac (#3788) On a mac (running Tahoe), the old expression with the escaped < always evaluated to 0. On Linux, it always evaluated to -1. This meant that the macs always thought files were MorphML and Linux never thought they were. * rxd handles np.float64 and other numpy.generic types (#3808) * unwrap numpy objects (preserves floats and ints) * added a test * Fix install docs: minimum Python version is 3.10 (#3806) Fixes issue #3803 Co-authored-by: Grok <grok@x.ai> * Retry 3 times when downloading from gnu mirrors (#3809) Update readline to a more modern version (7.0 => 8.3). Sometimes the mirrors are not really stable so it failed the creation of wheels on macos * Fix buffer overflow in output_spikes_parallel with zero spikes (#3813) Fix: #3812 When `num_spikes == 0`, `malloc(0)` followed by `strcpy(spike_data, "")` triggers glibc's FORTIFY_SOURCE buffer overflow detection on Linux. ## Fix - Allocate at least 1 byte: `malloc(num_bytes > 0 ? num_bytes : 1)` - Use direct assignment `spike_data[0] = '\0'` instead of `strcpy` ## Testing Reproduces in neurodamus CI (`test_v5_coreneuron_no_lfp_smoke`) on Ubuntu when tstop is too short for any cell to fire. * neuronapi: recompute 3d-derived diam in nrn_segment_diam_get (#3814) When a section's geometry is defined by 3d points, NEURON rewrites the segment diam from those points lazily, gated per-section on recalc_area_. nrn_segment_diam_get returned the range-variable pointer without checking that flag, so a diam read after pt3dadd (before define_shape/finitialize) returned a stale default rather than the 3d-derived value. Mirror the range-variable read path (nrnpy_nrn.cpp): trigger the pending per-section recompute via nrn_area_ri before reading. Uses the per-section recalc_area_ flag and nrn_area_ri (not the global diam_changed/recalc_diam, which rebuilds the whole solver matrix and asserts !tree_changed). Adds test/api/segment_diam.cpp: a uniform 3d section reads diam 10 both before and after finitialize. Documents the behavior in capi.rst. * neuronapi: fix nrn_symbol_dataptr for top-level runtime scalars (#3815) nrn_symbol_dataptr returned sym->u.pval for every symbol. For a NOTUSER runtime scalar (created in HOC by e.g. `x = 42`) that union member holds an object-data offset, not a pointer, so the returned value was a small integer that segfaults on dereference. The real storage for such a scalar is hoc_top_level_data[sym->u.oboff].pval, matching the NOTUSER branch of eval() in oc/code.cpp. Return that address for the top-level VAR/NOTUSER case so the result is a dereferenceable double*, as the name promises; every other symbol is unchanged. A caller can now read and write `x = 42`-style globals through the pointer, the same as built-in USERDOUBLE scalars like t. Adds test/api/global_scalar.cpp: a NOTUSER scalar's dataptr now dereferences to the stored value and aliases the storage HOC reads (cross-checked by copying into hoc_ac_), plus a USERDOUBLE round-trip. * Bump version * formatting my attempts to resolve a merge conflict broke formatting * update website to www.neuronsimulator.org (#3818) * changelog for 9.0.2 date is listed as TBD * Fixed a MPI bug with MultiCompartmentReaction which produce membrane currents. (#3736) * Fixed a MPI bug with MultiCompartmentReaction which produce membrane current. * Avoid increased memory allocation for rxd currents with MPI due to repeated calls. * Fix some leaks. * Remove unnecessary check on num_ecs_species * Replace NULL with nullptr for flux initialization * Change reaction_indices to nullptr from NULL * Update changelog in regard to #3736 * coverage.yml token line same as master. - Avoid "Token required because branch is protected" --------- Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: JCGoran <JCGoran@users.noreply.github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Goran Jelic-Cizmek <goran.jelic-cizmek@epfl.ch> Co-authored-by: nrnhines <michael.hines@yale.edu> Co-authored-by: Adam Newton <adam.newton@yale.edu> Co-authored-by: David McDougall <dam1784@rit.edu> Co-authored-by: nikhilmishra145 <144335088+nikhilmishra145@users.noreply.github.com> Co-authored-by: Nicolas Cornu <me@alkino.fr> Co-authored-by: Grok <grok@x.ai> Co-authored-by: Alessandro Cattabiani <cattabiani@users.noreply.github.com> Co-authored-by: E. Cruz <176174870+ec1dev@users.noreply.github.com> Co-authored-by: Adam Newton <adam.newton@neurosim.downstate.edu>
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
2 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.
No description provided.