| Age | Commit message (Collapse) | Author | |
|---|---|---|---|
| 2025-01-02 | smallcell: Reduce verbosity | Thomas White | |
| 2025-01-02 | indexer-choice.rst: Mention that smallcell has no dependencies | Thomas White | |
| 2025-01-02 | More small documentation updates | Thomas White | |
| 2025-01-02 | Update documentation about ffbidx and smallcell | Thomas White | |
| 2024-12-20 | Julia: Update image structure definition | Thomas White | |
| "id" was removed by 93e80b3f17c7a7ef226a92ae176b8eea7bbe5c79. | |||
| 2024-12-20 | Fix memory allocation routines | Thomas White | |
| 2024-12-20 | Julia: Use newer indexing routine prototype | Thomas White | |
| 2024-12-20 | Julia: Update CrystFEL.Indexing for FFBIDX | Thomas White | |
| 2024-12-20 | Merge branch 'smallcell' | Thomas White | |
| Closes: https://gitlab.desy.de/thomas.white/crystfel/-/merge_requests/35 | |||
| 2024-12-19 | Merge pull request #15 from hcstadler/master | Thomas White | |
| ffbidx indexer | |||
| 2024-12-19 | CrystFEL support for the ffbidx indexer (implementation of the TORO algorithm) | Hans-Christian Stadler | |
| Repository and description: https://github.com/paulscherrerinstitute/fast-feedback-indexer TORO algorithm paper: https://journals.iucr.org/j/issues/2024/04/00/jo5098/index.html Please cite the paper if you do research with this algorithm. Use --help-ffbidx to see options and --indexing=ffbidx to use it with indexamajig. | |||
| 2024-12-12 | indexamajig: Keep pumping data until there are no more complete chunks | Thomas White | |
| When things are running fast, multiple chunks can get sent through one pipe before the main process has a chance to write them to the unified output. However, the code would only check once. Now, it will keep trying until the pump routine returns zero. | |||
| 2024-12-12 | CI: Fix syntax | Thomas White | |
| 2024-12-12 | Shorten long set_last_task names | Thomas White | |
| 2024-12-12 | CI: Don't download all artifacts for deployment job | Thomas White | |
| The job only needs one version of the files. | |||
| 2024-12-12 | indexamajig: Add a load of set_last_task calls | Thomas White | |
| "indexing:finalisation" covers a lot of steps, and jobs seem to have been hanging there recently. This should help unblock the pipeline. | |||
| 2024-12-11 | indexamajig: Add final drain step for stream/Mille pipes | Thomas White | |
| 2024-12-11 | indexamajig: Add missing close of read end of pipe in main process | Thomas White | |
| Otherwise, the pipe never gets closed properly. | |||
| 2024-12-11 | indexamajig: Use poll() instead of select() | Thomas White | |
| 2024-12-11 | indexamajig: Exit if exec fails | Thomas White | |
| Otherwise, we might end up with multiple main processes! | |||
| 2024-12-11 | indexamajig: Clean up buffer read logic | Thomas White | |
| 2024-12-10 | indexamajig: Add a warning if closing a pipe with bytes remaining | Thomas White | |
| 2024-12-10 | indexamajig: Fix str_in_str | Thomas White | |
| It didn't properly detect the needle when at the end of the haystack, e.g. str_in_str("1234567890abc", 13, "abc") => NULL (should be "abc") | |||
| 2024-12-09 | GUI: Fix missing read of cell tolerances | Thomas White | |
| Previously, the tolerance values entered in the indexing options window were ignored! | |||
| 2024-11-14 | indexamajig: Avoid duplicate indexer info | Thomas White | |
| A consequence of fork-exec is that each worker process prints the indexer info at startup. We often have >100 processes, so that's a lot of spam. With this commit, the indexer info is printed by a separate routine which has to be called explicitly by the top level. Indexamajig only does this for worker 0. If worker 0 gets re-spawned, we'll see the message again. That seems to be an acceptable trade-off. The indexer info now won't get printed by the GUI, unless we go back and add the call to print_indexing_info. | |||
| 2024-11-14 | Remove image->id | Thomas White | |
| It wasn't used anywhere, and never got set to the correct value anyway. | |||
| 2024-11-08 | GUI: Add missing initialisation for peakfinder8_fast | Thomas White | |
| 2024-11-07 | CI: Move deployment stuff (e.g. Dockerfiles) to subfolder | Thomas White | |
| This is just to reduce clutter in the top level directory. | |||
| 2024-11-07 | Merge branch 'master' into 'master' | Thomas White | |
| Backup of MacOS runner configuration file. See merge request thomas.white/crystfel!42 | |||
| 2024-11-07 | Backup of MacOS runner configuration file. | Parthasarathy Tirumalai | |
| 2024-11-01 | Add article about choice of indexing method | Thomas White | |
| 2024-11-01 | README.md: Add link to DatView manual | Thomas White | |
| 2024-11-01 | CI: Change "Development Tools" to development-tools | Thomas White | |
| See https://github.com/rpm-software-management/dnf5/issues/1599 | |||
| 2024-11-01 | README.md: Add section on related software | Thomas White | |
| 2024-10-28 | peakfinder8: Disable max-res cutoff by default | Thomas White | |
| This changes the behaviour of peakfinder8 such that --max-res=0 disables the high-resolution cutoff completely. It then sets this value as the default for indexamajig and in the GUI. When peakfinder8 was first created, 1200 pixels was a generous cutoff. These days, we have many detectors that are much bigger and the default value for --max-res seems to create confusion. | |||
| 2024-10-28 | all_panels_perpendicular_to_beam: Fix units problem | Thomas White | |
| fsz and ssz are in units of pixels (of clen) per pixel (of pixel index). Therefore, the difference calculation result is in pixels. The limit is 10 pixels (of whatever panel). The old version was therefore far too conservative with what constitutes "flat" | |||
| 2024-10-28 | Clarify that coffset/cnz_offset is in metres | Thomas White | |
| 2024-10-28 | indexamajig: Set up peakfinder8 after fork/exec | Thomas White | |
| This restores the speed gains made by not calculating the radial maps for every frame. | |||
| 2024-10-22 | Add u1 format for Seedee | Thomas White | |
| 2024-10-22 | indexamajig: Fix incorrect peak list handling for MsgPack | Thomas White | |
| 2024-10-22 | indexamajig: Fix incorrect bounds check in pump_mille | Thomas White | |
| 2024-10-18 | CI: Fix insidious syntax problem that was disabling MsgPack | Thomas White | |
| 2024-10-18 | Fix memory leak in powder_rings() | Thomas White | |
| 2024-10-18 | smallcell: Fix memory leaks | Thomas White | |
| 2024-10-18 | smallcell: Simplify fit_cell | Thomas White | |
| 2024-10-18 | smallcell: Keep list of biggest cliques | Thomas White | |
| Instead of complaining when we run out of space, replace the smallest clique with the latest one (but only if the latest one is bigger). | |||
| 2024-10-18 | smallcell: Add missing check for right-handed cell | Thomas White | |
| 2024-10-18 | smallcell: Replace list manipulation with qsort | Thomas White | |
| 2024-10-18 | smallcell: Move find_max_cliques into separate function | Thomas White | |
| 2024-10-17 | smallcell: Tweak debugging messages | Thomas White | |
 |
index : CrystFEL | |
| Data processing for serial crystallography | Thomas White |
| aboutsummaryrefslogtreecommitdiff |