{"payload":{"feedbackUrl":"https://github.com/orgs/community/discussions/53140","repo":{"id":45596393,"defaultBranch":"master","name":"odl","ownerLogin":"odlgroup","currentUserCanPush":false,"isFork":false,"isEmpty":false,"createdAt":"2015-11-05T07:57:55.000Z","ownerAvatar":"https://avatars.githubusercontent.com/u/15667171?v=4","public":true,"private":false,"isOrgOwned":true},"refInfo":{"name":"","listCacheKey":"v0:1599376785.0","currentOid":""},"activityList":{"items":[{"before":"9a5b9d70ac27285a5715be83d3af359695547fa0","after":"60a854a8391d62addbbe1c617e076668ef14d523","ref":"refs/heads/master","pushedAt":"2024-03-08T09:47:07.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"ozanoktem","name":"Ozan Öktem","path":"/ozanoktem","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/439645?s=80&v=4"},"commit":{"message":"Create custom COO matrix representation for product operators, with updated product spaces (#1642)\n\n* create custom COO matrix representation for product operators\r\n\r\n* run linter\r\n\r\n* add documentation to coo matrix\r\n\r\n* fix linter issues\r\n\r\n---------\r\n\r\nCo-authored-by: Paul Hausner ","shortMessageHtmlLink":"Create custom COO matrix representation for product operators, with u…"}},{"before":"823eea3cfd0d2b581f39d0cf86305fc41214356b","after":"9a5b9d70ac27285a5715be83d3af359695547fa0","ref":"refs/heads/master","pushedAt":"2024-03-07T09:16:41.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"JevgenijaAksjonova","name":"Jevgenija Rudzusika","path":"/JevgenijaAksjonova","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9403836?s=80&v=4"},"commit":{"message":"Product space unit tests compatible with modern NumPy (#1640)\n\n* An array representation for testing product spaces with modern NumPy.\r\n\r\nNumPy does not really do ragged arrays anymore, unless dtype=object is used - in which case however\r\nnumerical operations are largely unsupported.\r\nThis solution is a compromise. It does not work perfectly, but is usable with some special-casing\r\nto carry out the unit tests.\r\n\r\n* Almost-equality check that works with the new array representations for product spaces.\r\n\r\n* Rewrite of the tests for real- and imaginary setters in product spaces.\r\n\r\nThe old ones explicitly checked on shapes (with a quite dubious special case for ℝ³).\r\nThis does not work anymore with the current reference arrays and more restrictive NumPy.\r\nThe new version uses something of a brute force attempt: we try to replicate the expected\r\nresult to match any product space structure. Not pretty, but it avoids hard-codes special\r\ncases for specific spaces and works for the present test cases.\r\n\r\n* Mangle code layout to satisfy pedantic linter rules.\r\n\r\nUnfortunately reduces readability a lot, but this is how the rules are.\r\n\r\n* Document the purpose of recursion_limit in pspace_test.","shortMessageHtmlLink":"Product space unit tests compatible with modern NumPy (#1640)"}},{"before":"c2c56d7dba262eaf2a78aeb0ec75b41c8fe028d6","after":"823eea3cfd0d2b581f39d0cf86305fc41214356b","ref":"refs/heads/master","pushedAt":"2024-02-22T10:00:06.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"JevgenijaAksjonova","name":"Jevgenija Rudzusika","path":"/JevgenijaAksjonova","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9403836?s=80&v=4"},"commit":{"message":"Compatibility with matplotlib-3.8. (#1638)\n\nThe update methods for colorbars have been removed in favour of more global updates.\r\n\r\nApplying the change to the `clim`s simply on the `csub` was suggested by paulhausner\r\nin https://github.com/odlgroup/odl/pull/1635 and seems to work fine.\r\n\r\nActually updating the drawing of the colorbar (specifically, the ticks on it) did\r\nnot work with the fix suggested in the Matplotlib documentation, but `canvas.draw_idle()`\r\ndoes seem to have the intended effect.\r\n\r\nFixes https://github.com/odlgroup/odl/issues/1636.","shortMessageHtmlLink":"Compatibility with matplotlib-3.8. (#1638)"}},{"before":"d4a5ed6efc391dcb61d65ffbcd79448b78681c38","after":"c2c56d7dba262eaf2a78aeb0ec75b41c8fe028d6","ref":"refs/heads/master","pushedAt":"2024-02-13T11:21:15.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"JevgenijaAksjonova","name":"Jevgenija Rudzusika","path":"/JevgenijaAksjonova","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9403836?s=80&v=4"},"commit":{"message":"Small compatibility fixes. (#1633)\n\n* Compatibility with scikit-image>=0.17.\r\n\r\n* Compatibility with numpy>=1.24.\r\n\r\nOlder versions would automatically convert the meshgrid used for scikit-image Radon transform\r\nto a ragged array, i.e. array of arrays. Newer versions do not do this anymore and need to\r\nbe explicitly asked for it.","shortMessageHtmlLink":"Small compatibility fixes. (#1633)"}},{"before":"139e7967e85927f9d1dd39da3942c04035eb81e3","after":"d4a5ed6efc391dcb61d65ffbcd79448b78681c38","ref":"refs/heads/master","pushedAt":"2024-02-05T21:22:16.000Z","pushType":"pr_merge","commitsCount":1,"pusher":{"login":"JevgenijaAksjonova","name":"Jevgenija Rudzusika","path":"/JevgenijaAksjonova","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/9403836?s=80&v=4"},"commit":{"message":"Fix a memory leak in the Operator base class (#1632)\n\n* Fix a memory leak in the Operator base class\r\n\r\nReferences to operator *class* definitions were cached in the argument cache of\r\n`_dispatch_call_args` function, hence GC couldn't collect them. Normally that's not\r\nan issue, as class definitions don't consume much memory. However, a typical\r\npattern of defining adjoint operators in ODL is using nested classes with\r\nreferences back to their encapsulating *objects*. Depending on the Operator, it\r\nmay cause considerable memory leaks. For example, `RayTransform` objects hold\r\nreferences to space and geometry objects, preventing them from being reaped by\r\nGC and leading to considerable memory leaks.\r\n\r\nThe fix I've implemented is moving away from an argument cache in favor of caching\r\nthe returned values on the class itself. This ensures no class object references are\r\nheld permanently in the cache_arguments() decorator.","shortMessageHtmlLink":"Fix a memory leak in the Operator base class (#1632)"}},{"before":"0b088df8dc4621c68b9414c3deff9127f4c4f11d","after":"139e7967e85927f9d1dd39da3942c04035eb81e3","ref":"refs/heads/master","pushedAt":"2023-08-08T08:16:28.000Z","pushType":"pr_merge","commitsCount":2,"pusher":{"login":"ozanoktem","name":"Ozan Öktem","path":"/ozanoktem","primaryAvatarUrl":"https://avatars.githubusercontent.com/u/439645?s=80&v=4"},"commit":{"message":"Merge pull request #1627 from JevgenijaAksjonova/numpy_version_issue_1623\n\ndeprecated types np.float and np.object substituted by float and object","shortMessageHtmlLink":"Merge pull request #1627 from JevgenijaAksjonova/numpy_version_issue_…"}}],"hasNextPage":false,"hasPreviousPage":false,"activityType":"all","actor":null,"timePeriod":"all","sort":"DESC","perPage":30,"cursor":"djE6ks8AAAAED_8mAwA","startCursor":null,"endCursor":null}},"title":"Activity · odlgroup/odl"}