129 Commits

Author SHA1 Message Date
d2fdf03d19 not working dump 2025-04-16 12:14:33 +02:00
643c53c894 Objects: less string concats 2025-04-15 17:07:03 +02:00
29fdd3eb08 Objects: don't calculate bundle size all the time from scratch 2025-04-15 17:02:26 +02:00
e6ead10e7f Objects: use direct bytebuffers when reading 2025-04-15 16:35:31 +02:00
04c5685fd5 Use @Singleton instead of @ApplicationScoped in hot paths
Definitely a microoptimization but noticeable on the flamegraph
2025-04-15 16:29:49 +02:00
67852fb37e Objects: less awful put, with 2 less copies 2025-04-15 11:17:23 +02:00
45556f2b74 Objects: rename name to value in JObjectKey 2025-04-13 17:10:33 +02:00
146870c281 Objects: support parallel test running
crashes with env maxreaders reached though
2025-04-13 14:54:25 +02:00
7c605135c5 Some native image support
but it still doesn't work because of JNR
2025-04-13 13:09:36 +02:00
491afd454b Objects: serialize written objects in parallel 2025-04-13 12:08:40 +02:00
a4810c7ee4 Objects: fix cache reading twice 2025-04-12 20:49:29 +02:00
e42e076b77 Objects: disable dhfs.objects.lru.print-stats by default
quite annoying
2025-04-09 11:02:46 +02:00
cb8c50000a Objects: simplify merging iterator even more^2
initialMaxValue streams can be simplified too
2025-04-03 22:31:59 +02:00
4c5cbfb5bf Objects: simplify merging iterator even more
no need for the hashmap step of iterator construction
2025-04-03 22:23:23 +02:00
6bcec4a260 Objects: simplify merging iterator
remove the first match "optimization", as it doesn't really
matter with the separate read object methods
2025-04-03 22:13:34 +02:00
df00584367 Objects: getting begin/end iterators 2025-04-03 15:05:09 +02:00
ea4f041d6e Objects: remove outdated snapshot stuff 2025-04-03 13:02:55 +02:00
3c37638db2 Objects: less locks in writeback 2025-04-02 16:14:50 +02:00
0e12a59f23 Objects: a little lazier caching 2025-04-02 15:35:02 +02:00
735dd605d7 Objects: get rid of locks in lmdb store 2025-04-02 15:13:58 +02:00
b998871e7f Objects: make cache status print debug 2025-03-30 22:08:00 +02:00
8b3c0a6f2c Server: snapshot interface key fix 2025-03-29 12:29:28 +01:00
8fdbaf5aa7 Objects: move objects to com.usatiuk.objects 2025-03-27 15:41:16 +01:00
4c5fd91050 Server: some refcounting fixes 2025-03-23 22:32:32 +01:00
c977b5f6c9 Run code cleanup 2025-03-23 15:29:32 +01:00
c5a875c27f Server: separate ref types
as we need to get the real "parent" object when sending some requests
2025-03-23 15:29:06 +01:00
8a07f37566 Server: a bunch of fixes 2025-03-22 21:31:20 +01:00
16eb1d28d9 Objects: add a "never lock option" to avoid deadlocks 2025-03-21 23:43:29 +01:00
6a20550353 Server: better logging and (hopefully) fix ktree op forwarding messing up timestamps 2025-03-21 20:23:59 +01:00
7d762c70fa Server: seemingly almost working file conflict resolution
though the removeAddHostTest doesn't seem to quite like it
2025-03-20 22:26:21 +01:00
97c0f002fb Objects: don't run tx commit hooks inside the same transaction 2025-03-17 23:29:10 +01:00
842bd49246 Objects: lmdb iterator find first fix 2025-03-16 00:08:05 +01:00
1b0af6e883 Objects: remove IterProdFn close() 2025-03-16 00:07:16 +01:00
667f8b3b42 Objects: a bunch of moving around 2025-03-14 22:39:54 +01:00
0aca2c5dbb Objects: move serializer stuff a little 2025-03-14 22:31:45 +01:00
223ba20418 Objects: run tx commit callbacks in empty transactions 2025-03-14 22:23:36 +01:00
6e37320e7c Objects: seemingly more reasonably working tx hooks 2025-03-14 16:21:07 +01:00
d37dc944d0 Objects: remove findAllObjects 2025-03-14 15:03:11 +01:00
d483eba20d Objects: working cache for object reads without iterator 2025-03-14 00:01:21 +01:00
5f85e944e3 Objects: fix read only transactions locking everything 2025-03-13 23:36:34 +01:00
38ab6de85b Objects: fix tests not being run by maven 2025-03-13 21:15:53 +01:00
29fd2826a3 Objects: simplify snapshots 2025-03-13 19:35:19 +01:00
3470ce8690 Objects: lazier iterators 2025-03-13 10:07:49 +01:00
1d22465e4a Objects: don't recreate iterators when initializing in MergingKvIterator 2025-03-12 21:21:10 +01:00
5c905379ae Objects: hopefully fix the file write deadlock
as some other random files could get into read set,
and a bit of refactoring
2025-03-12 12:10:24 +01:00
66bf4b7d18 Objects: really lazy deserialization
as it was happening in the cache
2025-03-08 15:27:08 +01:00
174320d389 Objects: lazy deserialization 2025-03-08 15:05:57 +01:00
6e14e05c4a cache lmdb transactions 2025-03-08 14:19:23 +01:00
51a6ba10d0 fix cache race
can't use linkedhashmap concurrently like that
2025-03-08 14:19:09 +01:00
75fec73b39 reverse iterators and a bunch of fixes with lt/gt iterator start 2025-03-08 00:35:56 +01:00