@rorydavidson -
The UK edition uses the following file names for descriptions and language reference set -
sct2_Description_Snapshot-en-GB_GB1000000_20181031.txt
The notation en-GB is not incorrect, but only used by the UK. This caused the files to be skipped on import.

The RF2 files within an Alpha release are prefixed "xsct2_" or "xder2_". Loading of these files should be supported.

when i test with LoadingProfile.complete config to load SnomedCT_RF2Release_INT_20160731 data,
i got an IllegalStateException.
then i found the code in ConceptImpl.java

private Set collectParentIds(ConceptImpl concept, Set ancestors, Stack stack) {
for (Concept parentInt : concept.parents) {
ConceptImpl parent = (ConceptImpl) parentInt;
final Long parentId = parent.id;
if (stack.contains(parentId)) {
stack.push(parentId);
throw new IllegalStateException("Ancestor loop detected: " + stack.toString());
}
ancestors.add(parentId);
stack.push(parentId);
collectParentIds(parent, ancestors, stack);
stack.pop();
}
return ancestors;
}

could you talk about why throws IllegalStateException there?
why not use continue?
thanks

The concept file is loaded in the main thread. If there is an issue loading the file, for example a missing file header, an exception is thrown from the ReleaseImporter class. However all other files are loaded in threads and any loading errors do not cause an exception to be thrown. This means that there is no way for an implementation to know if the complete content has been loaded or not.

The factory method names "createConcept", "addRelationship" and "addDescription" do not accurately reflect the meaning of a row in an RF2 file. It may not be a new concept being created but a new state of an existing concept. Similarly for relationships and descriptions, the new state of these components could be updating or retiring and existing component.
Names like "newConceptState" and "newDescriptionState" would be much more accurate and not lead to incorrect assumptions by developers using or extending the library.

Increase unit test coverage to at least 60% and publish stats on project readme page.
Key classes like ReleaseImporter should be covered first.

I have a use case where I would like to use snomed-boot to parse a RF2 delta zip file to just record the concept ids changed. Loading a delta is not currently possible.

For example the following file can not be loaded from an RF2 snapshot zip "sct2_Concept_SpanishExtensionSnapshot_INT_20180430.txt".

This type of naming is supported in the Snomed Release File specification so should be supported by this library https://confluence.ihtsdotools.org/display/DOCRELFMT/3.3.2+Release+File+Naming+Convention#id-3.3.2ReleaseFileNamingConvention-ContentSubTypeElement

I am seeing this error if the text definition RF2 file is missing from the archive I am loading:
2018-02-13 10:09:05,893 [ERROR ] [pool-1-thread-4] org.ihtsdo.otf.snomedboot.ReleaseImporter$ImportRun - Failed to read or process lines. java.lang.NullPointerException at java.nio.file.Files.provider(Files.java:97) at java.nio.file.Files.newInputStream(Files.java:152) at java.nio.file.Files.newBufferedReader(Files.java:2784) at org.ihtsdo.otf.snomedboot.ReleaseImporter$ImportRun.readLines(ReleaseImporter.java:412) at org.ihtsdo.otf.snomedboot.ReleaseImporter$ImportRun.access$200(ReleaseImporter.java:106) at org.ihtsdo.otf.snomedboot.ReleaseImporter$ImportRun$6.call(ReleaseImporter.java:363) at org.ihtsdo.otf.snomedboot.ReleaseImporter$ImportRun$6.call(ReleaseImporter.java:359) at java.util.concurrent.FutureTask.run(FutureTask.java:266) at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149) at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624) at java.lang.Thread.run(Thread.java:748)

Tried this with the Swedish release and it seems it was not designed to handle extensions. Is national releases within scope of this package?
Cheers,
Daniel

the fix for issue #3 introduces if statement that is an impossible to satify breaking all relation information.
see ReleaseImporter line 229

It would really speed things up if we could filter reference sets by filename when loading.

I've just been caught out loading a large RF2 reference set which had no header row. SNOMED Boot loaded the file without any error. It told me that the field names of the reference set were the values from the first row which was actually an OWL expression value!

The library should check that the first line of every RF2 file being loaded starts with "id effectiveTime active moduleId".

Only the inferred relationships should be used to calculate the transitive closure.