11534 Commits (0632ef464736ca7ad59abe8c1b8ac7c24d454e6e)

 

All Branches   

Search

Author	SHA1	Message	Date
Rich Harris	0632ef4647	Merge branch 'async-another-try-pt-1' into async-another-try-pt-1-prune	2 days ago
Rich Harris	c66fab19b4	prettier	2 days ago
Rich Harris	cea4875a2f	add tests from #18983 that fail on main but pass here	2 days ago
Rich Harris	7f4460e16e	Merge branch 'async-another-try-pt-1' into async-another-try-pt-1-prune	2 days ago
Rich Harris	4e314973d3	Merge branch 'main' into async-another-try-pt-1	2 days ago
Rich Harris	86eb09b76e	chore: even more async tests (#18990) ... extracting some tests from #18983 — these pass on `main`, `pt-1`, `pt-2` and `pt-3`. The others in that PR don't pass on `main`, so will be dealt with separately	2 days ago
Rich Harris	69f0bb549f	restore resetting effects to CLEAN in is_dirty while time travelling ... Effects left MAYBE_DIRTY are forced to re-run when a paused branch is resumed.	2 days ago
Rich Harris	336d1ef608	restore first_batch?.next check when assigning to a derived in internal_set ... Marking the derived clean while other batches are in flight hides changes they haven't seen yet.	2 days ago
Rich Harris	fa16f59447	restore capturing deriveds that match the real value but not the batch's view ... Otherwise the batch keeps seeing another batch's previous value for the derived.	2 days ago
Rich Harris	737de2e4f3	restore removal of batch_values.delete(derived) in mark_reactions ... A derived's batch-local value can be a later batch's previous value; deleting it makes the next read fall back to the real (later batch's) value.	2 days ago
Rich Harris	70bb7f4295	restore capturing deriveds in previous_batch before current_batch ... The committing batch needs the correct previous value, and the new batch must record that its value came from the committing batch's world.	2 days ago
Rich Harris	498a696ac8	restore updating content.wv on fork commit ... apply() uses the batch's recorded write versions while other batches are pending, so they must match the fresh versions given to the sources on commit.	2 days ago
Rich Harris	3e92edfbf4	restore transferring discard callbacks in #merge ... If the surviving batch is discarded (e.g. after an unhandled error), the merged batch's discard callbacks must run too, or each blocks keep its pending state and leak items.	2 days ago
Rich Harris	652103bb9f	restore is_eager early return in #find_earlier_batch ... An eager batch sees other batches' previous values, so its async results must not resolve an earlier batch's work. It can still get a dependency via the in-flight check in async_derived.	2 days ago
Rich Harris	7a0b430ec5	restore stale_readers dedupe ... Without it, a reaction that reads N stale values queues N re-runs, each of which flushes the later batch — quadratic work (seconds for a 2000-item each block). The new test is a timing guard: the output is identical either way.	2 days ago
Rich Harris	2db9eccefe	Merge branch 'main' into async-another-try-pt-1	2 days ago
Rich Harris	4d2ad8a189	chore: add async-merge-blocking-pending test from #18983 (#18989) ... #18983 adds two new tests. One of them passes on `main`, so we can merge it independently. The other only passes on `pt-1`, so I'll add it to that branch	2 days ago
Rich Harris	4fef105eaa	Merge branch 'async-another-try-pt-1' into async-another-try-pt-1-prune	2 days ago
Nic	5d7eef91f0	simplify the two tests	2 days ago
Nic	90dc9fc329	test: cover branch commit order and merged blocking pending	2 days ago
Rich Harris	a029f0b756	Merge branch 'async-another-try-pt-1' into async-another-try-pt-1-prune	2 days ago
Rich Harris	d69a3c3c1c	Update packages/svelte/src/internal/client/reactivity/batch.js	2 days ago
Rich Harris	1df42fe0c4	unnecessary	2 days ago
Rich Harris	95f5c157db	restore capturing writes in internal_set that match the real value but not the batch's view ... Needed on async-another-try-pt-2 (async-fork-write-back).	2 days ago
Rich Harris	deace209ff	restore removal of !is_flushing_sync check in Batch.ensure ... Needed on async-another-try-pt-3 (async-fork-branch-merge).	2 days ago
Rich Harris	cc8b5d7089	restore stale_effects ... Needed on async-another-try-pt-2/pt-3: recording stale effects, the wv override in apply(), and marking them maybe-dirty in later batches are all required by fork tests there (e.g. async-fork-effect-write-version, async-state-new-branch-fork-5).	2 days ago
Simon Holthausen	a8f5384326	Merge branch 'main' into async-another-try-pt-1	2 days ago
Rich Harris	d9acd2da90	test: add async tests that already pass on main, unskip async-overlap-multiple-2 (#18981) ... These tests from #18880 already pass on `main`. Moving them out of that PR makes it clearer what behaviour is actually changing over there	2 days ago
Nic	f60f99e6d8	format	3 days ago
Nic	f2dd258299	simplify the two tests	3 days ago
Nic	00fbd15467	restore the batch id check in BranchManager#commit and carry blocking_pending through #merge	3 days ago
Nic	28fb668d48	test: cover branch commit order and merged blocking pending	3 days ago
Rich Harris	8ab778b621	merge	3 days ago
Rich Harris	fd3ddfb933	chore: don't transfer #blocking_pending and discard callbacks in #merge ... Neither is needed for any test, unlike the #pending transfer — worth a closer look.	3 days ago
Rich Harris	1d6ef08ccf	chore: remove small pieces not needed by any test ... - stale-source dependency in get_batch_value - saving/restoring stale_sources in eager() - is_eager early return in #find_earlier_batch - updating content.wv on fork commit	3 days ago
Rich Harris	5eb0051d4b	chore: simplify the later-batch check in capture ... The check for this batch's own dependency values differing from the real ones isn't needed for any test.	3 days ago
Rich Harris	3d25df0cb4	chore: don't remove dependency-less deriveds from #dirty_reactions ... Not needed for any test. The call for clean deriveds is still needed (async-derived-deferred-clean).	3 days ago
Rich Harris	e0cb3c667c	chore: don't reset effects to CLEAN in is_dirty while time travelling ... Not needed for any test.	3 days ago
Rich Harris	927e5baff2	chore: remove first_batch?.next checks for reads outside reactivity ... Neither the check in internal_set nor the one in update_derived is needed for any test.	3 days ago
Rich Harris	579224846e	chore: don't capture writes that match the real value but not the batch's view ... Neither the else-if branch in internal_set nor its equivalent in update_derived is needed for any test.	3 days ago
Rich Harris	b60c1b23de	chore: remove stale_readers ... The dedupe isn't needed for any test, including async-stale-read-multiple-no-double-run.	3 days ago
Rich Harris	4d65c92b96	chore: remove stale_effects ... Neither the wv override in apply() nor marking the effect maybe-dirty in later batches is needed for any test — only not bumping the effect's write version is.	3 days ago
Rich Harris	8ec850dd12	chore: keep main's code where the branch's change isn't needed ... - record block effects as maybe-dirty during traversal - drop the extra DIRTY guard in #process - delete deriveds from batch_values in mark_reactions - don't queue a microtask flush in Batch.ensure during flushSync - BranchManager commit loop - capture order in update_derived	3 days ago
Rich Harris	8a7499d3cc	merge main	3 days ago
Rich Harris	95db764ce5	merge main	3 days ago
Rich Harris	6480a9737f	chore: make Batch prev/next/started public (#18982) ... converts some private fields to public ones, because on #18880 they are used outside the class. breaking the change out removes some diff noise	3 days ago
Rich Harris	0ecfc6079c	merge main	3 days ago
Rich Harris	e22a67c432	chore: tidy up comments and types in batch.js and sources.js (#18980) ... couple of trivial tidy-ups, ref #18880	3 days ago
Rich Harris	397051dd85	merge main	3 days ago
Rich Harris	0c61f1bb76	chore: merge #dirty_effects and #maybe_dirty_effects into #dirty_reactions (#18979) ... yoinking a minor change out of #18880	3 days ago