Jan Niestadt
                                            Jan Niestadt
                                        
                                    The method used to manually advance systemList.head and call system.removeFromEngine(), but removeSystem() already takes care of that. This eventually resulted in a null reference. The offending lines were removed, and...
Useful while debugging to detect if an entity that should have been removed is still in the engine, or if an active component is somehow still in the ComponentPool (maybe...
By default, Lucene will merge segment files into a single compound file to use fewer file descriptors. This takes a bit of time (7-33% during indexing according to [docs](https://cwiki.apache.org/confluence/display/lucene/ImproveIndexingSpeed)) and...
If all unique terms combined total more than 2GB of character data, TermsReader will break. See [TermReader](https://github.com/INL/BlackLab/blob/evolve-hits-interfaces/engine/src/main/java/nl/inl/blacklab/forwardindex/TermsReader.java#L218): ```java // FIXME this code breaks when char term data total more than...
Right now a value is sent, but it is never updated if we add or remove documents from an existing corpus. If we properly keep track of this, we can...
Some hits occur twice, due to indexing both lemma and word in one annotation. The setting `allowDuplicateValues: false` should have prevented this. https://corpusoudnederlands.ivdnt.org/corpus-frontend/ONL/search/hits?first=0&number=20&patt=%5Bword_or_lemma%3D%22Ne%22%5D%5Bword_or_lemma%3D%22willen%22%5D&interface=%7B%22form%22%3A%22search%22%2C%22patternMode%22%3A%22simple%22%7D
Right now, multithreading works per-file. This means that a single large file that contains many documents cannot use more than one CPU core. We could try reading a few input...
Certain (rare) inputs can cause this error. We're trying to set a payload for a non-existent position. How can this happen? Could this be an empty document containing only a...
There are still files with mixed tabs and spaces around. This tends to cause commit noise, inconsistent indenting in diffs, etc. We should consider switching the import style to using...
See comment in colloc.js (on branch better-integration-tests)