aboutsummaryrefslogtreecommitdiff
path: root/libcrystfel
AgeCommit message (Collapse)Author
2020-10-15TakeTwo: Limit number of seeds to 10, and reduce MAX_OBS_VECTORS to 300Thomas White
This gives a huge speed improvement, without sacrificing very much in indexing rate.
2020-10-07Revert "Move asdf out of set of indexers used by default"Thomas White
This reverts commit 62183aacf906f4aff771295aa378cff039dd50ff. The apparent slowness was due to a bug introduced in August 2019. See 2a8f415df845064186baa833f9fe1ad91c0f5753.
2020-10-07asdf: Fix incorrect units conversion for cell volumeThomas White
2020-10-07Formatting fussinessThomas White
2020-10-07Improve indexer speed profilingThomas White
2020-10-07FussinessThomas White
Commented out debug message is no longer relevant
2020-10-07TakeTwo: Set tp->optsThomas White
2020-10-07TakeTwo: Fix get_chiral_holohedry for "hexagonal H"Thomas White
Fixes CRYS-240
2020-10-06Move asdf out of set of indexers used by defaultThomas White
Sadly, it has the same fatal flaw as TakeTwo: taking forever on a minority of patterns.
2020-10-01asdf: Replace weird loop syntaxThomas White
2020-09-30image_expand_frames: Report non-existent file properlyThomas White
2020-09-25stream_make_index: s/Indexing/Scanning/Thomas White
Because it sounds strange and confusing to index the indexing results.
2020-09-24Fix old warning messageThomas White
2020-09-24Apply default unit scale factor only when value is from headersThomas White
Ugh.
2020-09-17CMake: Change LIBRARIES to LDFLAGSThomas White
Necessary after conversion to pkg-config search, to make sure that custom paths get used.
2020-09-17Add missing includeThomas White
2020-09-17Ensure that fs dimension has higher index than ssThomas White
Reading from HDF5 cannot rearrange dimensions, only select a chunk. Therefore, the layout in memory needs to match the file.
2020-09-16image_hdf5_read: Don't call image_freeThomas White
You didn't allocate it, so you don't free it.
2020-09-16Fix incorrect logic in HDF5 expansionThomas White
2020-09-11xgandalf: Remove C99-style 'for' declarationsThomas White
2020-09-10CMake: Don't compile image-msgpack.c if library is not foundThomas White
2020-09-09get_matrix_name: Improve string handlingThomas White
2020-09-09Add optional measurement of time spent in indexersThomas White
2020-09-08Remove "Command line:" in streamThomas White
This wasn't present in the old version, and it was missing a space anyway.
2020-09-08Un-export image_read_image_dataThomas White
Not needed since f5b3c4db36fa.
2020-09-07Catch another "NULL" event -> //Thomas White
2020-09-04stream_read_chunk: Set image->spectrumThomas White
2020-09-04stream_make_index: Don't require an event ID for every chunkThomas White
Event ID is not written if not needed
2020-09-04make_key: Handle NULL eventThomas White
2020-09-04New API for random access to streamThomas White
2020-09-02Get rid of STREAM_IMAGE_DATAThomas White
No longer needed, and didn't fit very well in the first place.
2020-08-27TakeTwo: Remove internal declarationsThomas White
2020-08-27Fix indentationThomas White
2020-08-27Only compile image-msgpack.c if MessagePack is foundThomas White
2020-08-25Factor out convert_intThomas White
2020-08-20Add default_method_options()Thomas White
This allows the indexing method private options structures to be set up easily.
2020-08-19Set the same initial beam parameters as beforeThomas White
2020-08-19Update old commentsThomas White
2020-08-19Remove STREAM_UNITCELL and STREAM_CRYSTALSThomas White
These flags weren't orthogonal at all, which was turning some routines into spaghetti-code. It's still possible to avoid read/write of peak lists, reflection data or image data. But that's it.
2020-08-19Const-cleaningThomas White
2020-08-19Make Stream responsible for its own DataTemplateThomas White
It makes no sense for a different DataTemplate to be used for every frame in a stream. And this way makes everything much easier for users ot the API.
2020-08-18Formatting/commentsThomas White
2020-08-18predict-refine: Remove out-of-bounds GSL vector accessThomas White
We got away with this because the GSL error handler is turned off in indexamjig.c, and invalid accesses are defined to return 0. However, these accesses should have been removed when the clen refinement was disabled in predict-refine
2020-08-17Expose parse_indexing_methods and base_indexer_strThomas White
2020-08-17FormattingThomas White
2020-08-11image_read_image_data: Set zero values if file not foundThomas White
2020-08-10image_read: Tolerate event==NULLThomas White
2020-08-10Disable pinkIndexerThomas White
Temporary, of course. But it needs some thought about how the interface should work.
2020-08-10Start on converting pinkIndexer to DataTemplateThomas White
2020-08-10pinkIndexer: Remove detector updateThomas White
Overall detector offsets can now be done via the geometry file, and this is labelled as a hack. It doesn't actually do very much anyway, because the "updated" detector is thrown away for the next pattern. The updated detector is used only for subsequent indexing attempts on the same pattern.