89 commits

Author	SHA1	Message	Date
ppom	f1f4c2ea2c	WIP custom Time ... Permitting custom de/serialiazing	2025-06-03 12:00:00 +02:00
ppom	ebad317a97	Config: fix state_directory and concurrency merging	2025-05-28 12:00:00 +02:00
ppom	9152c95b03	Change error wording again	2025-05-28 12:00:00 +02:00
ppom	f7bd28e46e	Config: cleanup code	2025-05-28 12:00:00 +02:00
ppom	73ffcb97ab	Update CLI help messages	2025-05-28 12:00:00 +02:00
ppom	1a5548c871	Reorder Config fields for test-config command ... This order looks more readable to me, with: - one line options at the beginning of config - streams at the end of config - filters at the end of streams - actions at the end of filters	2025-05-28 12:00:00 +02:00
ppom	4f2eac2788	Change error wording	2025-05-28 12:00:00 +02:00
Baptiste Careil	231c9f8a99	Add test-config sub-command	2025-05-28 12:00:00 +02:00
Baptiste Careil	28b3a173bb	Add ability to read config from multiple files in a same directory	2025-05-28 12:00:00 +02:00
ppom	5260b6b9b1	Clean handle_order ... Had bits that were relevant in LMDB but are not anymore. Use of keys().cloned() instead of custom map() Clone after filtering.	2025-05-28 12:00:00 +02:00
ppom	a5bbd7641a	reaction {show,flush}: do not display empty pending actions ... We were creating an action array even if it was past.	2025-05-28 12:00:00 +02:00
ppom	3ffb2b5bed	Adapt to Time being serialized as rfc3339 ... And all tests pass! 🎉	2025-05-28 12:00:00 +02:00
ppom	3d05815263	Daemon runs the database task ... Completely forgot to run the database task, no wonder why the test didn't pass 😅 Now the database asks for the daemon to stop if it errors. A lot better than panicking, as it was until now. It also quits with all the other tasks, thanks to its ShutdownToken. It uses a oneshot channel to tell the daemon about a potential error.	2025-05-28 12:00:00 +02:00
ppom	78ca0025df	Document daemon::shutdown. ShutdownToken is now able to ask for shutdown.	2025-05-28 12:00:00 +02:00
ppom	0c56acfd81	Fix FilterManager unit tests ... e2e test failing for now	2025-05-28 12:00:00 +02:00
ppom	1783bf8062	Use treedb in FilterManager ... Not tested yet, tests must be adapted	2025-05-28 12:00:00 +02:00
ppom	037b3498bc	rename waltree into treedb ... WAL was a wrong name. It's not a Write Ahead Log, but a "Write Behind Log" (new concept haha), so it made no sense to keep wal. And wbl is not unpronounceable.	2025-05-28 12:00:00 +02:00
ppom	5f21db5279	WriteDB: reuse write buffer	2025-05-28 12:00:00 +02:00
ppom	fe5cd70f7a	use tokio::time::interval instead of sleep, for guaranteed flushes	2025-05-28 12:00:00 +02:00
ppom	c93f4e34c1	LoadedDB internal to Database, open_tree uses LoadedDB. More doc!	2025-05-28 12:00:00 +02:00
ppom	482df254ba	implement Tree::fetch_update	2025-05-28 12:00:00 +02:00
ppom	ec0d8c72e9	Reorganise waltree. Test log rotation. Add tokio channel. Implement Tree operations.	2025-05-28 12:00:00 +02:00
ppom	6cdad37588	Finish tests on lowlevel db (and fix bug)	2025-05-28 12:00:00 +02:00
ppom	924c8e8635	test ReadDB::read	2025-05-28 12:00:00 +02:00
ppom	a029e28812	Tested ReadDB and WriteDB	2025-05-28 12:00:00 +02:00
ppom	779e5e5d86	more WIP, split into low level / high level	2025-05-28 12:00:00 +02:00
ppom	bb9b17761a	WIP custom db	2025-05-28 12:00:00 +02:00
ppom	9c8be2f2de	Working db rewrite using fjall	2025-05-28 12:00:00 +02:00
ppom	2facac9fbd	WIP fjall	2025-05-28 12:00:00 +02:00
ppom	11459f7ee4	lmdb NO_SYNC	2025-05-28 12:00:00 +02:00
ppom	db0e480daa	arbitrary max size ... To avoid a runtime Error(MapSize) by heed FIXME How to know which value is high enough?	2025-05-28 12:00:00 +02:00
ppom	a056efe770	Fix warning until next heed release	2025-05-28 12:00:00 +02:00
ppom	d952a61f83	Fix last lmdb bug ... Hopefully the last. Tests pass! I was using a read transaction outdated by recent changes in an uncommited write transaction. So I use the same write transaction to read things now. I also checked that other uses of a read transaction is fine. They're fine because they're only used to iterate on a whole db, while the write transaction drops some keys/values.	2025-05-28 12:00:00 +02:00
ppom	64fdc52e9a	Better use of transactions, React enum for more precise tests, test macros ... Transactions are now opened at higher level, to 1. Ensure a logical transaction is a complete action. (ex: when exec, remove match & add corresponding trigger) 2. Fix issue where multiple Rw / Ro transactions are open at the same time. So now they're open only by: - FilterManager's task in handle_*, - FilterManager's task in cleanup_*, - the scheduled task for after actions. handle_line returns an enum (no match, match, exec) to have more precise tests. So maybe the tests could care less about database state and only about FilterManager's behavior. Use of macros to have more readable tests and reduce duplication and remove complex generic functions Tests still failing, behavior seems to be altered.	2025-05-28 12:00:00 +02:00
ppom	ea1bc033c5	WIP	2025-05-28 12:00:00 +02:00
ppom	8393ebf20e	Temporary fix while we can't remove heed databases ... Also remove sled related code	2025-05-28 12:00:00 +02:00
ppom	77f001f860	Add debug log before closing DB	2025-05-28 12:00:00 +02:00
ppom	e321e3ea0b	collecting maps before any remove_match to avoid iterators holding a read ... But now the tasks don't finish so I introduced another bug?	2025-05-28 12:00:00 +02:00
ppom	04e3fb3e28	unit tests passing. handle_order in e2e tests panicing ... panic because of two read transactions open in same thread. Don't know if I will fix this with txns without tls or if i'll find another way	2025-05-28 12:00:00 +02:00
ppom	3b6c352204	WIP	2025-05-28 12:00:00 +02:00
ppom	0abb55b69e	WIP	2025-05-28 12:00:00 +02:00
ppom	9e83ceed46	WIP	2025-05-28 12:00:00 +02:00
ppom	660a7d5a58	WIP	2025-05-28 12:00:00 +02:00
ppom	a297f26f3d	Prettier exec_limit handler	2025-05-03 12:00:00 +02:00
ppom	94502443f7	Prettier pattern-matching 💅	2025-04-29 12:00:00 +02:00
ppom	f4b8572e94	run clippy on test target	2025-02-26 12:00:00 +01:00
ppom	b655ef1008	Move tests, add tests, fix memory leak ... - Move FilterManager tests to own file - Add some persistance tests - Fix memory leak where entry like (Match, EmptySet) would be kept in DB (instead of discarding it)	2025-02-26 12:00:00 +01:00
ppom	b448089f58	Added daemon tests; parse_duration now supports milliseconds ... TODO test persistance handling on FilterManager	2025-02-25 12:00:00 +01:00
ppom	fe1a93b8a2	Remove more dead code ... Client/Daemon communication is now JSON so we don't need This quirks anymore	2025-02-23 12:00:00 +01:00
ppom	fb9c0f8699	Remove dead code ... Process is a bit tedious, as pub objects are not treated as unused: 1. Remove all `pub`s: sed -i 's/\bpub //' (rg -le rs src/) 2. Fix the errors in your IDE by adding the necessary `pub`s 3. Remove what's marked as unused, when the project compiles again.	2025-02-23 12:00:00 +01:00