|
From: Lars W. <no...@gi...> - 2026-10-14 11:29:00
|
Branch: refs/heads/develop Home: https://github.com/eXist-db/exist Commit: 8d946a3b59ba050b942a2eb46e65add4de443885 https://github.com/eXist-db/exist/commit/8d946a3b59ba050b942a2eb46e65add4de443885 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/org/exist/test/xqsuite/XQSuite.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteSchedulingTest.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteSettings.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngine.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngineTest.java Log Message: ----------- [test] Skip the XQSuite engine's own fixture suites unless a test asks for them The fixture suites of the engine tests are nested static classes annotated with XQSuite. A run that selects them by name (a -Dtest wildcard such as XQSuiteSchedulingTest*, or the package or classpath scan of an IDE) ran them as ordinary suites. The hanging fixture then took the default five minutes to be given up on, and the run reported two failures for each of SequentialHang and ParallelHang, as well as the deliberately failing files of XQSuiteTestEngineTest. XQSuite gets a fixture flag. The engine skips such a suite unless the configuration parameter exist.xqsuite.fixtures is true, which the engine tests set on their EngineTestKit runs. A guard test checks that a fixture is skipped without the parameter and fails without the engine change. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Commit: 31d65173682b1253cd24856eda12fcfa1848adf2 https://github.com/eXist-db/exist/commit/31d65173682b1253cd24856eda12fcfa1848adf2 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/org/exist/test/xqsuite/SuiteRun.java A exist-core/src/test/java/org/exist/test/xqsuite/SuiteRunAwaitTerminationTest.java Log Message: ----------- [bugfix] Bound the wait for abandoned XQSuite threads by one grace period in all A test file that hangs in a database lock is given up on, but its thread does not end when it is interrupted. SuiteRun then waits for the abandoned threads before the database is shut down, and computed the time left for each one as max(0, deadline - now) for Thread.join. The first stuck thread uses up the whole grace period, and join(0) does not mean "do not wait", it means "wait forever", so with two or more stuck files the run never ended. The same happens for any thread when the grace period is set to 0. The wait is now a helper that skips the threads once the time is spent, and returns whether all of them ended. The tests use threads that ignore interruption, and fail (time out) with the previous formula. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Commit: 529deb25bc05360d34eb6d59a5a673436d232aea https://github.com/eXist-db/exist/commit/529deb25bc05360d34eb6d59a5a673436d232aea Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/org/exist/test/xqsuite/SuiteRun.java A exist-core/src/test/java/org/exist/test/xqsuite/SuiteRunTimeSummaryTest.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngineTest.java Log Message: ----------- [test] Report the run time of the XQSuite files Surefire has no per-file entry. Publish the time of each file as a report entry, and print one line per suite when it is done, with the number of files, the time in all and the slowest files. A line printed as each file ends would land in the output of whichever test is running at that moment, which with files running side by side is a test of another file. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 0443081c272e235a6440d913e31fe1deb9c6236e https://github.com/eXist-db/exist/commit/0443081c272e235a6440d913e31fe1deb9c6236e Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/main/java/org/exist/xslt/XsltURIResolverHelper.java M exist-core/src/test/xquery/transform/catalog.xql Log Message: ----------- [bugfix] XSLT: try the system catalog before the database resolver EXistURIResolver opened absolute http URLs without a connect timeout before the system catalog was consulted. transform/catalog.xql imports from the non-routable address 203.0.113.1 and expects the catalog to redirect the import, so every run of XQuery3Tests waited for the OS connect timeout: 150 s, with a thread dump at 53 s showing the thread inside the resolver's connect. XQuery3Tests went from 184 s to 8 s. The resolvers are now tried in this order: EXPath package resolver, system catalog, database resolver, default resolver. A catalog entry still wins over a live network fetch (see #350). A product fix that does not depend on the test isolation work. Also drops an unused import from XsltURIResolverHelper. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 7a73327c4c0841c959531e3c821bf161522ac2ea https://github.com/eXist-db/exist/commit/7a73327c4c0841c959531e3c821bf161522ac2ea Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/xquery/axes-persistent-nodes.xqm M exist-core/src/test/xquery/axes.xml M exist-core/src/test/xquery/guest.xql M exist-core/src/test/xquery/json.xml M exist-core/src/test/xquery/locationStepVariables.xml M exist-core/src/test/xquery/npt.xqm M exist-core/src/test/xquery/parenthesizedContext.xml M exist-core/src/test/xquery/parenthesizedLocationStep.xml M exist-core/src/test/xquery/pathExpression_operators.xml M exist-core/src/test/xquery/persistent_in-memory.xml M exist-core/src/test/xquery/question-marks-in-contains-function.xql M exist-core/src/test/xquery/range.xml M exist-core/src/test/xquery/rewriting.xml M exist-core/src/test/xquery/selfAxis.xml Log Message: ----------- [test] Give CoreTests files unique collection names Files shared /db/coll, /db/test and /db/axes-test and removed each other's data when run in parallel. Each file now uses a collection named after itself (for example test-guest, test-npt, test-axes-persistent-nodes). Setup, teardown and every reference are renamed together; no paths are rewritten in the harness. Serial runs are unchanged. axes.xml now removes its collection in its tearDown again: the removal was commented out (it still named the old collection), unlike in the other XSuite files; nothing else refers to /db/axes-xml-test. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: e5282373032865796f7302599bad29b869802ce6 https://github.com/eXist-db/exist/commit/e5282373032865796f7302599bad29b869802ce6 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/org/exist/test/xqsuite/XQSuite.java M exist-core/src/test/java/xquery/CoreTests.java Log Message: ----------- [test] Run the CoreTests files in parallel 12 of 12 parallel runs green at 2, 4 and 8 files at a time; the suite takes 5.1 s serial and 2.6-3.6 s parallel. The rules the files must follow are in the Javadoc of XQSuite#parallel(), which this commit extends. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: d8e8a3242a8e6eb9f539fd5e6dba94d5566892f0 https://github.com/eXist-db/exist/commit/d8e8a3242a8e6eb9f539fd5e6dba94d5566892f0 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/xquery/xquery3/annotations.xml M exist-core/src/test/xquery/xquery3/forwhere.xml M exist-core/src/test/xquery/xquery3/higher-order.xqm M exist-core/src/test/xquery/xquery3/inline.xml M exist-core/src/test/xquery/xquery3/recursion_calls.xml M exist-core/src/test/xquery/xquery3/xq3_switch.xml M exist-core/src/test/xquery/xquery3/xq3_trycatch.xml Log Message: ----------- [test] Give XQuery3Tests files unique collection names Files shared /db/xq3-test and /db/xquery3, so two of them running at the same time stored and removed each other's documents. Each file now uses a collection named after itself (for example xq3-higher-order). Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 60ba728760f467610b7dacb278591dcda5271e3d https://github.com/eXist-db/exist/commit/60ba728760f467610b7dacb278591dcda5271e3d Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/xquery/xquery3/XQuery3Tests.java Log Message: ----------- [test] Run the XQuery3Tests files in parallel 12 of 12 parallel runs green at 2, 4 and 8 files at a time; the suite takes 8.1 s serial and 5.1 s parallel. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: dd849a8c6676622347e832f8e9c391b701b3da36 https://github.com/eXist-db/exist/commit/dd849a8c6676622347e832f8e9c391b701b3da36 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/xquery/logicalOp/and.xml M exist-core/src/test/xquery/logicalOp/or.xml M exist-core/src/test/xquery/optimizer/expressions.xqm M exist-core/src/test/xquery/optimizer/positional.xqm M exist-core/src/test/xquery/system/get-main-module-load-path.xqm M exist-core/src/test/xquery/system/get-module-load-path.xqm Log Message: ----------- [test] Give LogicalOp, System and Optimizer files unique collections and.xml and or.xml, the two module-load-path files of the system suite and the two optimizer files shared a collection each, so two of them running at the same time would store and remove each other's documents. Each file now uses a collection named after itself (for example test-module-load-path, optimizertest-expressions). Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 7d9abbd8177f5280a8f6bfd38d2168c8776d59c0 https://github.com/eXist-db/exist/commit/7d9abbd8177f5280a8f6bfd38d2168c8776d59c0 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/indexes/range/src/test/xquery/range/optimizer.xql Log Message: ----------- [test] Give optimizer.xql its own copy of the range.xql collection optimizer.xql used range.xql's collection range-test-range, with the same index configuration and data, for its tests of nested placeName fields, so the two files running at the same time shared documents. It now creates its own copy, range-optimizer-copy. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 832db6a77ce1078209a3c1a488d38ca3fb232330 https://github.com/eXist-db/exist/commit/832db6a77ce1078209a3c1a488d38ca3fb232330 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/modules/file/src/test/resources/util/fixtures.xqm M extensions/modules/file/src/test/resources/util/helper.xqm M extensions/modules/file/src/test/xquery/modules/file/sync-serialize.xqm M extensions/modules/file/src/test/xquery/modules/file/sync.xqm M extensions/modules/file/src/test/xquery/modules/file/syncmod.xqm Log Message: ----------- [test] Give the FileTests sync files unique collections The three sync files (sync, sync-serialize, syncmod) shared the collection that the shared fixtures hard-coded, /db/file-module-test. Each file now declares its own (/db/file-module-test-sync, -sync-serialize and -syncmod) and passes it to helper:setup-db and helper:clear-db, which take the collection as a parameter; the collection variables of the fixtures are removed. Also removes helper:sync-with-options, a wrapper around file:sync that nothing calls, and the two sync files' unused child-collection variables. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 208d316ff66a807d20a05ca30afab9af83349f4b https://github.com/eXist-db/exist/commit/208d316ff66a807d20a05ca30afab9af83349f4b Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/modules/xmldiff/src/test/xquery/modules/xmldiff/diff.xqm Log Message: ----------- [test] Give XmlDiffTests diff.xqm its own collection diff.xqm and compare.xqm used the same collection (test-xmldiff-compare), so running at the same time they removed each other's documents. diff.xqm now uses test-xmldiff-diff. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 040d3d914e9e2c598c687c214afae0cdd89c5830 https://github.com/eXist-db/exist/commit/040d3d914e9e2c598c687c214afae0cdd89c5830 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/org/exist/test/runner/ExtTestFailureFunction.java A exist-core/src/test/java/org/exist/test/runner/ExtTestFailureFunctionTest.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngineTest.java Log Message: ----------- [test] Show failed values in XQSuite reports and cap huge ones A failed XQSuite assertion reached the surefire report as one line, for example "XQuery failure: ft-query-field.xqm:352 issue5431-sorted-by-binary-field". The expected and the actual value were carried by the AssertionFailedError, but surefire keeps only the message and the stack trace, so a wrong result could not be told from an empty one without running the test again by hand. * Build the failure with Jupiter's AssertionFailureBuilder, which puts "==> expected: <x> but was: <y>" into the message and still carries the values for an IDE comparison. The second "at (file:line)" line of the message is gone; the headline already names the file and line, and the stack trace keeps them. * Cut an expected or actual value at 4000 characters, saying how long it was, so that one very large result cannot flood the report. The same limit applies to the values an IDE shows for comparison. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 0ca719e3403cd5cc39a4c39bb8b9eb42b667a1bb https://github.com/eXist-db/exist/commit/0ca719e3403cd5cc39a4c39bb8b9eb42b667a1bb Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/org/exist/test/runner/AbstractTestRunner.java M exist-core/src/test/java/org/exist/test/runner/XQueryTestRunner.java M exist-core/src/test/java/org/exist/test/xqsuite/FileDescriptor.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngine.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngineTest.java M exist-core/src/test/resources/org/exist/test/runner/xquery-discovery.xq Log Message: ----------- [test] Name the test file in XQSuite reports and link the test source Surefire reports all tests of a suite under the suite class and uses the display name of a test as its name, so tests of different files were indistinguishable in a report, and equal names in two files collided. * The display name of a test now leads with its file, for example "ft-query-field.xqm: issue5431-sorted-by-binary-field". The report names of XQSuite tests change. * Give each test the line its function is declared on, as the position of its file source. Before, an IDE opened the test file at the top; now it jumps to the test. The line comes from util:inspect-function in the discovery query. XML tests, and tests found by compiling the module, have no line and keep the file only. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 1daa5ce9e3f6de4f5863c1f1a0eca7a76eeec9db https://github.com/eXist-db/exist/commit/1daa5ce9e3f6de4f5863c1f1a0eca7a76eeec9db Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/indexes/lucene/src/main/java/org/exist/indexing/lucene/LuceneIndexWorker.java A extensions/indexes/lucene/src/test/java/org/exist/indexing/lucene/LuceneBinaryFieldLookupTest.java Log Message: ----------- [bugfix] Lucene: read the binary field from the reader the search ran on ft:binary-field returned an empty value for some nodes, which showed as a flaky issue5431-sorted-by-binary-field in LuceneIndexingTests when its files ran side by side. LuceneIndexWorker.getBinaryFieldByExistDocId had two defects: * It found the Lucene document number with a search on the searcher, and then read the value through a second, separately acquired reader. The searcher and the reader refresh independently, and a number is only valid in the snapshot it was found in, so a commit by another thread in between could point the lookup at another document. The value is now read from the reader of the searcher that found the number. * It compared the number within a segment with the number of live documents (numDocs) instead of all documents (maxDoc). A segment with deleted documents, which a reindex or the deletion of another document leaves, has more numbers than live documents, so the documents numbered above the count were not found. The segment is now found with ReaderUtil.subIndex and the number is used as it is. The new test stores a document that is then deleted before one that is kept, so that the kept one is numbered above the live documents of the segment, and reads the binary field of every item of it. It fails on the unfixed code (item 012 and the two after it have no value) and checks that its precondition holds, so it cannot pass without testing the lookup. With the fix, 40 runs of LuceneIndexingTests with its files running side by side (8 at a time) had no failure; before, about 1 in 5 runs failed. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 824ce9e1d633722ce03dd486529014b8690211e0 https://github.com/eXist-db/exist/commit/824ce9e1d633722ce03dd486529014b8690211e0 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/indexes/lucene/src/test/xquery/lucene/ft-query-forms.xqm Log Message: ----------- [test] Bind the index-keys-by-qname Lucene tests to their own collection index-keys-by-qname-all and index-keys-by-qname-s (and the pending attribute-type test) called util:index-keys-by-qname without a context, which scans the terms of every document in the database and compares them with the terms of this file's own documents. Any other file that indexed a p element at the same time added terms, so the result depended on which other files were running. They now run on the documents of the file's own collection, like the test for a collection that does not exist beside them. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 2179ca55db26cfbee9feef31fa94d8ecb82e0549 https://github.com/eXist-db/exist/commit/2179ca55db26cfbee9feef31fa94d8ecb82e0549 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/indexes/lucene/src/main/java/org/exist/indexing/lucene/LuceneIndexWorker.java A extensions/indexes/lucene/src/test/java/org/exist/indexing/lucene/LuceneIndexScanSparseDocValuesTest.java Log Message: ----------- [bugfix] Give each term of an index scan its own doc values iterators LuceneIndexWorker.doScanIndex (util:index-keys and util:index-scan) read the doc id and the node id of each hit from doc values iterators that it fetched once per segment. The hits of every term start again at the first document, but a doc values iterator only moves forward: advanceExact needs a target at or above the current one. A segment in which every document has the value tolerates going back; a segment in which only some do, which happens once documents without a node id (written by ft:index) share it with indexed nodes, does not. With assertions on, as in tests, the scan failed with an AssertionError in IndexedDISI.advanceExactWithinBlock; with assertions off it can return wrong values. It showed in LuceneTests when files ran side by side (index-scan-path-lucene, index-scan-qname-lucene and their -has-terms tests). Each term now gets its own iterators. The new test writes a Lucene document without a node id into the segment of three indexed items, checks that the segment is in that state, and scans terms whose hits start before the last hit of the term before. It fails on the unfixed code with the AssertionError. Also drop the FIXME that said the doc id iterator was null and should not be: both places that write documents add the doc id doc values, and the stored field fallback is kept for older indexes. This was not verified at run time. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 1a9fcc999a1fb34084f0eab9650b1971bcc9e065 https://github.com/eXist-db/exist/commit/1a9fcc999a1fb34084f0eab9650b1971bcc9e065 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/indexes/lucene/src/test/xquery/lucene/ft-search-binary.xqm Log Message: ----------- [test] Make ft:search tests independent of score order and other files search-title-text and search-title-text-db returned the right two documents in the other order when other files indexed documents at the same time: ft:search orders its hits by score, and the score depends on the statistics of the whole index (document frequencies and the average field length), which other files change. The tests check which documents match, so the uris are sorted, in these two and in the two other tests of the file that expect two uris. search-title-text-db searches the whole database ("/db/") on purpose, so it now keeps only the hits in its own collection and no longer fails on the documents of other files. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 4d98dc756443c0d18936d1983bac3f50e055cdef https://github.com/eXist-db/exist/commit/4d98dc756443c0d18936d1983bac3f50e055cdef Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/indexes/lucene/src/main/java/org/exist/indexing/lucene/LuceneIndex.java M extensions/indexes/lucene/src/main/java/org/exist/indexing/lucene/LuceneIndexWorker.java A extensions/indexes/lucene/src/main/java/org/exist/indexing/lucene/analyzers/FieldAnalyzerWrapper.java A extensions/indexes/lucene/src/test/java/org/exist/indexing/lucene/analyzers/FieldAnalyzerWrapperTest.java Log Message: ----------- [bugfix] Lucene: analyze with the analyzer of the adding thread The index and its IndexWriter are shared by all collections, and the analyzer of a field is looked up by the name of the field, which does not tell collections apart: every collection that indexes p elements writes into the same field, with the analyzer of its own index configuration. LuceneIndexWorker.write() put the analyzer into one shared map and then called addDocument, but the writer analyzes the document when addDocument gets to it, and addDocument can wait for a flush that another thread started. If a thread that indexes into another collection registered its own analyzer for the same field in the meantime, the document was analyzed with that one: with a collection that does not fold diacritics next to one that does, "Ruesselsheim" with an umlaut was indexed unfolded next to the folded term, and a stemming English collection lost its stemming. The index writer now uses FieldAnalyzerWrapper, which takes the analyzers a thread sets for the duration of its own addDocument call (LuceneIndex.addDocument) before it looks into the shared map. The shared map stays as the registry for everything else, and is now a ConcurrentHashMap, because it is written by several threads. Found with analyzers-diacritics.xql and analyzers-english.xql, which index p with different analyzers and run side by side at 2 files at a time: the pair failed in 16 of 74 runs without the change and in 0 of 20 with it. The swap itself happens inside addDocument and could not be observed, so FieldAnalyzerWrapperTest models it: it registers an analyzer from another thread while a document is being added. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 67f26bd870dd2dda196e26687e953462f37a3af1 https://github.com/eXist-db/exist/commit/67f26bd870dd2dda196e26687e953462f37a3af1 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/indexes/lucene/src/main/java/org/exist/indexing/lucene/LuceneIndexWorker.java A extensions/indexes/lucene/src/test/java/org/exist/indexing/lucene/LuceneFacetsReaderLifecycleTest.java Log Message: ----------- [bugfix] Lucene: keep the taxonomy reader alive while facets are read The facets of a search (ft:facets) are computed from the taxonomy reader of the searcher the query ran on, but that searcher is handed back to the manager as soon as the query returns. The index is shared by all collections, so a write by another thread followed by a refresh replaces the searcher, and the old taxonomy reader was closed: reading the facets then failed with "this TaxonomyReader is closed". Serial runs never refresh between the query and ft:facets, so only parallel runs (and a server with concurrent writers) saw it. The facets now hold a reference on their taxonomy reader, which a Cleaner releases once the facets are unreachable. LuceneFacetsReaderLifecycleTest reads the facets of a search after another document was stored and the index refreshed; it fails with the error above without this change. Also drops three unused imports from LuceneIndexWorker. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: c8b01eaecb2cc24c1a0d3fcbb1de5eb60955cc17 https://github.com/eXist-db/exist/commit/c8b01eaecb2cc24c1a0d3fcbb1de5eb60955cc17 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/main/java/org/exist/xquery/Profiler.java M exist-core/src/main/java/org/exist/xquery/XQueryContext.java A exist-core/src/main/java/org/exist/xquery/functions/system/QueryTrace.java M exist-core/src/main/java/org/exist/xquery/functions/system/SystemModule.java M exist-core/src/main/resources/org/exist/xquery/lib/xqsuite/xqsuite.xql A exist-core/src/test/java/org/exist/xquery/functions/system/QueryTraceTest.java M exist-core/src/test/xquery/xqsuite/xqsuite-tests.xql Log Message: ----------- [feature] Add query-scoped trace functions for the XQSuite stats tests system:enable-tracing, system:trace and system:clear-trace work on the statistics of the whole database instance. While tracing is on, every query records into them, and a query that reads or clears them reads or clears what all the others recorded. A %test:stats test that runs next to another test file therefore sees calls it did not make, or an empty trace after the other file cleared it, and the test files of a suite cannot run in parallel. The new functions system:enable-query-tracing, system:query-trace and system:clear-query-trace use the statistics that every query keeps for itself (imported modules and evaluated queries share them with the query that loaded them), never touch the database-wide ones, and do not write the global tracelog setting. The three existing functions are unchanged, so monitoring that reads the database-wide trace is not affected. The XQSuite harness uses the new functions for %test:stats, and switches tracing off when a test throws: the call that normally does it is never reached then, and tracing stayed on for the rest of the file's query. QueryTraceTest runs two queries at the same time and checks that a query that never switched tracing on records nothing while another has it on, that the database-wide flag stays off, and that clearing one query's trace keeps another's. Two tests in xqsuite-tests.xql check that nothing is recorded after a stats test that threw. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 304e988be2812b842202ee48f2497a24ca655866 https://github.com/eXist-db/exist/commit/304e988be2812b842202ee48f2497a24ca655866 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/indexes/range/src/test/java/xquery/range/RangeTests.java M extensions/indexes/range/src/test/xquery/range/conditions.xql M extensions/indexes/range/src/test/xquery/range/index-keys.xql M extensions/indexes/range/src/test/xquery/range/prefixed-conditions.xql M extensions/indexes/range/src/test/xquery/range/range.xql Log Message: ----------- [test] Run the RangeTests files in parallel, with scoped queries The files of this suite used the database-wide performance trace for their %test:stats tests, so parallel runs failed in the files that check the index statistics. With the query-scoped trace functions that is gone: 24 of 24 runs were green at 2, 4 and 8 files at a time (the suite takes about 4 s either way, so the point is to keep it independent, not to save time). Parallel runs also exposed queries that were not scoped to a file's own collection. On the Windows CI job range.xql update-delete failed once, after more than 30 green local runs: six calls of range:field-eq without a collection context found the copy of the same data that optimizer.xql stores, because both files index the field address-name and store the same names. The calls are now collection($rt:COLLECTION)//range:field-eq(...). range.xql also creates a bystander collection with the same index and documents, as another file would, so a query that is not scoped fails here every time and not only when another file happens to run at the same moment: without the scoping update-delete, update-replace and update-value fail with the bystander present. range:index-keys-for-field() called without a context reads every document of the database. The 17 calls in conditions.xql and prefixed-conditions.xql were safe only because each field name is defined in exactly one file; they are scoped as well, and conditions.xql has a bystander with the same index and other key values: with the calls unscoped 5 tests fail (index-ends-with, index-eq-numeric, index-lt-numeric, index-lt-non-numeric, index-starts-with); scoped, the suite is 411 tests, 0 failures, 3 of 3 runs at 8 files at a time. The two tests of index-keys.xql that call without a context on purpose, and the two wrapper functions there, are left as they are, with a comment. Also drops the unused variable $rt:DATA2 of range.xql. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 4d1a8d9b6984c25302b30bfd4b6ecedb0ab885ea https://github.com/eXist-db/exist/commit/4d1a8d9b6984c25302b30bfd4b6ecedb0ab885ea Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/xquery/optimizer/OptimizerTests.java Log Message: ----------- [test] Run the OptimizerTests files in parallel The suite has two files, expressions.xqm and positional.xqm, each with its own collection since the unique-name commit. Its %test:stats tests used the database-wide trace, which is gone with the query-scoped trace functions: 12 of 12 runs were green at 2, 4 and 8 files at a time (60 tests each; the suite takes about 4 s either way, so the point is to keep it independent, not to save time). Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 3da38efdeac3709e579f407ad25d8768ae586e64 https://github.com/eXist-db/exist/commit/3da38efdeac3709e579f407ad25d8768ae586e64 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/main/java/org/exist/repo/ClasspathHelper.java M exist-core/src/main/java/org/exist/repo/Deployment.java M exist-core/src/main/java/org/exist/repo/ExistRepository.java A exist-core/src/test/java/org/exist/repo/ExistRepositoryConcurrencyTest.java M exist-services/src/main/java/org/exist/repo/PackageService.java M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/GetResource.java M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/InstallFunction.java M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/ListFunction.java M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/RemoveFunction.java M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/ResourceAvailable.java Log Message: ----------- [bugfix] Serialize EXPath package installs and removals All packages of a database share one registry: the package map of the wrapped Repository and the files packages.xml and packages.txt. The library (EXPath pkg-java 2.1.1) locks its package map only around parts of an install or removal; choosing the directory of the package, adding the version to its Packages and the rewrite of the registry files run outside that lock, and the files are rewritten in place (read, transform, truncate, write). Two installs or removals at the same time could read a half-written file ("Error transforming the file: .../packages.xml"), and could overwrite each other's change and drop a package from the registry. Any server with concurrent package installs could hit this; it is what kept the two ExpathRepoTests files that install packages from running in parallel. ExistRepository now has installPackage (from an XarSource or a URI) and removePackage (all versions or one) under a read-write lock, and the call sites (Deployment, PackageService, the repo:install and repo:remove functions) use them instead of the wrapped Repository. The whole call is the critical section, not only the file write, because choosing the directory key and adding the version are also unlocked in the library. A package that does not come from a local file is first copied to a temporary file before the lock is taken, and installed from the copy through a small source that still reports the original address, so a slow download does not hold up other installs; the temporary file is deleted afterwards. A failure to read the source is reported as the repository reports it ("Error unziping the package" with the IOException, which carries the NotFoundException or HttpException of a download). installPackage(URI) delegates to the XarSource overload and keeps the read-only storage check. The library returns a live view of its package map from listPackages(), after releasing its own lock, so a caller that does work per package (the module lookups read files for each one) failed with a ConcurrentModificationException when a package was installed or removed during the iteration. ExistRepository.listPackages() returns a copy taken under the read lock, and the callers that iterate (ExistRepository, Deployment, ClasspathHelper, PackageService, repo:list, repo:get-resource, repo:resource-available) use it. getParentRepo() documents why not to use the repository's own list or to change the registry through it. ExistRepositoryConcurrencyTest (embedded server, 5 tests): six threads installing and removing packages of their own at the same moment (failed 2 of 2 on the unlocked code with the production error, green in every run with the lock); the registry on disk lists exactly the installed packages; a listing is not disturbed by an install during the iteration (forced interleaving; fails with ConcurrentModificationException on the repository's own list); a download that sends its headers and stalls does not hold up another install or a listing (fails after 30 s when the download is read under the lock, checked by disabling the copy); a failed download reports a PackageException that carries the NotFoundException (pins the existing error, does not fail without the change). Not covered: the Packages objects in a copy are the library's own, so the versions of one package can still change under a reader; two installs of the same package run one after the other and the second replaces the first (that is what force means). Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 193084d690853881180eda11a11c457404f99ad6 https://github.com/eXist-db/exist/commit/193084d690853881180eda11a11c457404f99ad6 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M extensions/modules/expathrepo/src/test/java/xquery/modules/expathrepo/ExpathRepoTests.java Log Message: ----------- [test] Run the ExpathRepoTests files in parallel The two files that install and remove packages (deployment.xql and resource-available.xql) failed together with "Error transforming the file: .../packages.xml": 2 of 4 runs at 2 files at a time. The package registry changes are serialized now (the two preceding commits), and 12 of 12 runs of the suite were green at 2, 4 and 8 files at a time, 9 tests each, on the code of this branch (the suite takes about 1 s either way, so the point is to keep it independent, not to save time). Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: 7663df7c1413b8a490637afb9af3b3e1872e0e78 https://github.com/eXist-db/exist/commit/7663df7c1413b8a490637afb9af3b3e1872e0e78 Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/org/exist/xquery/functions/fn/FunDeepEqualPerformanceTest.java Log Message: ----------- [test] Make the fn:deep-equal timing tests relative, not absolute FunDeepEqualPerformanceTest failed on earlier heads of this PR and could not be trusted: its three timing tests compared one run against a fixed number of milliseconds, and the 1500 ms cap of the root-mismatch test measured something other than what it claimed. That test builds two 10,000-element in-memory trees and compares them, and the comparison stops at the first element name, so it times the construction. On three successful CI runs (Ubuntu unit job) it took 1019, 1053 and 1025 ms, 68 percent of its cap, and longer than the equal-trees query that does the whole comparison (557 to 792 ms); locally the mismatch (48 ms), the equal trees (49 ms) and the construction alone (51 ms) are indistinguishable, so no threshold could have detected a lost short-circuit. The stored-docs guard for GH-4050 ran 1140, 1589 and 2712 ms on CI against a cap of 5000 ms and about 155 ms locally. Each test now compares two figures measured one after the other on the same machine, each the fastest of 5 runs after 2 warm-ups (noise only makes a run slower): - stored equal documents: at most twice a plain pass over the same two documents (count of nodes and attributes); about 0.29 times locally; - root mismatch: now on stored documents of the same size (a third document with another root name), where nothing has to be constructed, at most a tenth of the full comparison; 1.3 ms against 155 ms locally. The in-memory root mismatch keeps a correctness test without timing; - in-memory equal trees: at most 5 times the same query without the comparison; about 1.05 times locally. Validated by mutation, main code reverted afterwards: - streaming fast path disabled (FunDeepEqual.tryStreamingCompare returns null): the stored guard fails, 3171 ms against 521 ms for the pass, where the old cap of 5000 ms would have passed; - no short-circuit at a root name mismatch (both documents are read to the end before the mismatch is returned, result unchanged): the mismatch test fails, 110.9 ms against 160.3 ms, where the old cap of 1500 ms would have passed; - unmutated, all 15 tests pass in two runs. Not yet seen on CI, and not run under heavy machine load: the ratios on the CI runners are unmeasured (the earlier scratch measurements under 4 busy processes showed the same ratios for the same queries, but that was not these tests). Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Claude-Session: https://claude.ai/code/session_01CLyS76KK3RFSauL2JEqqiY Commit: a2d6a83066d22506f21ef568582666db58e52f0e https://github.com/eXist-db/exist/commit/a2d6a83066d22506f21ef568582666db58e52f0e Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/xquery/xmldb/XMLDBTests.java Log Message: ----------- [test] Run the XMLDBTests files in parallel The suite was left serial because xmldb:copy-collection deadlocked with another file's test when the files ran side by side: the copy took the read lock of the source Collection before the write lock of the target (the parent of the source), and the other tests held intention locks on /db. The lock order of copying and moving a Collection was fixed in #6775. The seven files use collection names of their own, so they do not disturb each other. Twelve runs on this branch (four reps each at parallelism 2, 4 and 8) passed 45 of 45 tests, none hung. Without the lock order fix the same configuration deadlocked in two of three runs, so the parallel suite also guards the lock order. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Commit: c464e462f46691e01bbbe6e3e765b1b57ad8974f https://github.com/eXist-db/exist/commit/c464e462f46691e01bbbe6e3e765b1b57ad8974f Author: Juri Leino <gi...@li...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/main/resources/org/exist/xquery/lib/xqsuite/xqsuite.xql M exist-core/src/test/java/org/exist/test/runner/ExtTestFailureFunction.java M exist-core/src/test/java/org/exist/test/runner/ExtTestFailureFunctionTest.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngineTest.java Log Message: ----------- [test] Keep XQSuite failure values clean for IDE comparison IntelliJ cut the message at "expected: <" and left a dangling "==>"; the assertion message was mixed into the expected value. Put the values on a line of their own and the assertion message in the headline. Co-Authored-By: Claude Opus 5.5 (1M context) <no...@an...> Commit: 0970500586046ddc36912c33d25babb87e0b592a https://github.com/eXist-db/exist/commit/0970500586046ddc36912c33d25babb87e0b592a Author: Juri Leino <gi...@li...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/org/exist/test/runner/ExtTestFailureFunction.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngineTest.java Log Message: ----------- [test] Name the suite and test in the XQSuite failure stack frame IDEs showed the frame as "at . (file.xq:70)". Co-Authored-By: Claude Opus 5.5 (1M context) <no...@an...> Commit: 24d8a7d7a035123c45fbf61204885c1af7e7534f https://github.com/eXist-db/exist/commit/24d8a7d7a035123c45fbf61204885c1af7e7534f Author: Juri Leino <gi...@li...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/main/resources/org/exist/xquery/lib/xqsuite/xqsuite.xql Log Message: ----------- [test] Keep test:failure-messages out of the xqsuite module's public API Co-Authored-By: Claude Opus 5.5 (1M context) <no...@an...> Commit: eb9d1b086eab48ee9dbcefe4fede5ac37156d0ee https://github.com/eXist-db/exist/commit/eb9d1b086eab48ee9dbcefe4fede5ac37156d0ee Author: Juri Leino <gi...@li...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/main/resources/org/exist/xquery/lib/xqsuite/xqsuite.xql M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngineTest.java A exist-core/src/test/resources/org/exist/test/runner/failing-boolean.xqm Log Message: ----------- [test] Report true and false as the expected values of assertTrue and assertFalse Co-Authored-By: Claude Opus 5.5 (1M context) <no...@an...> Commit: ed19d57101854e8265c270b6bceaa99a4767cd2c https://github.com/eXist-db/exist/commit/ed19d57101854e8265c270b6bceaa99a4767cd2c Author: duncdrum <d.p...@me...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/test/java/org/exist/xquery/functions/xmldb/DbStore2Test.java Log Message: ----------- [test] Bind the DbStore2Test server to a free port The mock HTTP server took a random port from 30350 to 45349, a range that overlaps the ephemeral ports of outgoing connections. When the picked port was in use, beforeClass failed with "Failed to bind to 0.0.0.0:38222" and the whole class errored, which failed the unit job of #6796 although the same commit passed in the sibling run. Bind to port 0 and read the port the OS assigned from the connector once the server has started. Co-Authored-By: Claude Sonnet 5.5 <no...@an...> Commit: 08211e94c83f518c4f6dba7a4fe2d4ccf5113993 https://github.com/eXist-db/exist/commit/08211e94c83f518c4f6dba7a4fe2d4ccf5113993 Author: Duncan Paterson <la...@ex...> Date: 2026-10-14 (Wed, 14 Oct 2026) Changed paths: M exist-core/src/main/java/org/exist/repo/ClasspathHelper.java M exist-core/src/main/java/org/exist/repo/Deployment.java M exist-core/src/main/java/org/exist/repo/ExistRepository.java M exist-core/src/main/java/org/exist/xquery/Profiler.java M exist-core/src/main/java/org/exist/xquery/XQueryContext.java A exist-core/src/main/java/org/exist/xquery/functions/system/QueryTrace.java M exist-core/src/main/java/org/exist/xquery/functions/system/SystemModule.java M exist-core/src/main/java/org/exist/xslt/XsltURIResolverHelper.java M exist-core/src/main/resources/org/exist/xquery/lib/xqsuite/xqsuite.xql A exist-core/src/test/java/org/exist/repo/ExistRepositoryConcurrencyTest.java M exist-core/src/test/java/org/exist/test/runner/AbstractTestRunner.java M exist-core/src/test/java/org/exist/test/runner/ExtTestFailureFunction.java A exist-core/src/test/java/org/exist/test/runner/ExtTestFailureFunctionTest.java M exist-core/src/test/java/org/exist/test/runner/XQueryTestRunner.java M exist-core/src/test/java/org/exist/test/xqsuite/FileDescriptor.java M exist-core/src/test/java/org/exist/test/xqsuite/SuiteRun.java A exist-core/src/test/java/org/exist/test/xqsuite/SuiteRunAwaitTerminationTest.java A exist-core/src/test/java/org/exist/test/xqsuite/SuiteRunTimeSummaryTest.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuite.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteSchedulingTest.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteSettings.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngine.java M exist-core/src/test/java/org/exist/test/xqsuite/XQSuiteTestEngineTest.java M exist-core/src/test/java/org/exist/xquery/functions/fn/FunDeepEqualPerformanceTest.java A exist-core/src/test/java/org/exist/xquery/functions/system/QueryTraceTest.java M exist-core/src/test/java/org/exist/xquery/functions/xmldb/DbStore2Test.java M exist-core/src/test/java/xquery/CoreTests.java M exist-core/src/test/java/xquery/optimizer/OptimizerTests.java M exist-core/src/test/java/xquery/xmldb/XMLDBTests.java M exist-core/src/test/java/xquery/xquery3/XQuery3Tests.java A exist-core/src/test/resources/org/exist/test/runner/failing-boolean.xqm M exist-core/src/test/resources/org/exist/test/runner/xquery-discovery.xq M exist-core/src/test/xquery/axes-persistent-nodes.xqm M exist-core/src/test/xquery/axes.xml M exist-core/src/test/xquery/guest.xql M exist-core/src/test/xquery/json.xml M exist-core/src/test/xquery/locationStepVariables.xml M exist-core/src/test/xquery/logicalOp/and.xml M exist-core/src/test/xquery/logicalOp/or.xml M exist-core/src/test/xquery/npt.xqm M exist-core/src/test/xquery/optimizer/expressions.xqm M exist-core/src/test/xquery/optimizer/positional.xqm M exist-core/src/test/xquery/parenthesizedContext.xml M exist-core/src/test/xquery/parenthesizedLocationStep.xml M exist-core/src/test/xquery/pathExpression_operators.xml M exist-core/src/test/xquery/persistent_in-memory.xml M exist-core/src/test/xquery/question-marks-in-contains-function.xql M exist-core/src/test/xquery/range.xml M exist-core/src/test/xquery/rewriting.xml M exist-core/src/test/xquery/selfAxis.xml M exist-core/src/test/xquery/system/get-main-module-load-path.xqm M exist-core/src/test/xquery/system/get-module-load-path.xqm M exist-core/src/test/xquery/transform/catalog.xql M exist-core/src/test/xquery/xqsuite/xqsuite-tests.xql M exist-core/src/test/xquery/xquery3/annotations.xml M exist-core/src/test/xquery/xquery3/forwhere.xml M exist-core/src/test/xquery/xquery3/higher-order.xqm M exist-core/src/test/xquery/xquery3/inline.xml M exist-core/src/test/xquery/xquery3/recursion_calls.xml M exist-core/src/test/xquery/xquery3/xq3_switch.xml M exist-core/src/test/xquery/xquery3/xq3_trycatch.xml M exist-services/src/main/java/org/exist/repo/PackageService.java M extensions/indexes/lucene/src/main/java/org/exist/indexing/lucene/LuceneIndex.java M extensions/indexes/lucene/src/main/java/org/exist/indexing/lucene/LuceneIndexWorker.java A extensions/indexes/lucene/src/main/java/org/exist/indexing/lucene/analyzers/FieldAnalyzerWrapper.java A extensions/indexes/lucene/src/test/java/org/exist/indexing/lucene/LuceneBinaryFieldLookupTest.java A extensions/indexes/lucene/src/test/java/org/exist/indexing/lucene/LuceneFacetsReaderLifecycleTest.java A extensions/indexes/lucene/src/test/java/org/exist/indexing/lucene/LuceneIndexScanSparseDocValuesTest.java A extensions/indexes/lucene/src/test/java/org/exist/indexing/lucene/analyzers/FieldAnalyzerWrapperTest.java M extensions/indexes/lucene/src/test/xquery/lucene/ft-query-forms.xqm M extensions/indexes/lucene/src/test/xquery/lucene/ft-search-binary.xqm M extensions/indexes/range/src/test/java/xquery/range/RangeTests.java M extensions/indexes/range/src/test/xquery/range/conditions.xql M extensions/indexes/range/src/test/xquery/range/index-keys.xql M extensions/indexes/range/src/test/xquery/range/optimizer.xql M extensions/indexes/range/src/test/xquery/range/prefixed-conditions.xql M extensions/indexes/range/src/test/xquery/range/range.xql M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/GetResource.java M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/InstallFunction.java M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/ListFunction.java M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/RemoveFunction.java M extensions/modules/expathrepo/src/main/java/org/exist/xquery/modules/expathrepo/ResourceAvailable.java M extensions/modules/expathrepo/src/test/java/xquery/modules/expathrepo/ExpathRepoTests.java M extensions/modules/file/src/test/resources/util/fixtures.xqm M extensions/modules/file/src/test/resources/util/helper.xqm M extensions/modules/file/src/test/xquery/modules/file/sync-serialize.xqm M extensions/modules/file/src/test/xquery/modules/file/sync.xqm M extensions/modules/file/src/test/xquery/modules/file/syncmod.xqm M extensions/modules/xmldiff/src/test/xquery/modules/xmldiff/diff.xqm Log Message: ----------- Merge pull request #6796 from eXist-db/test/xqsuite-isolation [feature] Isolate XQSuite test files, run CoreTests and XQuery3Tests in parallel, add system tracing functions Compare: https://github.com/eXist-db/exist/compare/d9b854559dd3...08211e94c83f To unsubscribe from these emails, change your notification settings at https://github.com/eXist-db/exist/settings/notifications |