docs: consigner deux défauts vivants, dont une absence qui masque un échec

This commit is contained in:
Sylvain Duchesne
2026-08-20 11:05:34 +02:00
parent c61b181ed8
commit 134cff3ad1
3 changed files with 49 additions and 7 deletions
-7
View File
@@ -1,7 +0,0 @@
# Doc-debt — app-contract
> Presence of a block = doc to update. Processed → delete the block; no blocks left → delete this file.
> One block = one "big change": `why` + `files` + `verify` (leaves to review).
## Raw markers (consolidate into blocks, then delete)
- TOUCHED packages/polyfill/src/surface/subscribe.ts @2026-08-20 (session f93872b5-293a-4916-a353-181409a96d42)
@@ -0,0 +1,28 @@
---
type: bug
severity: major
summary: An inbox watch that failed to open is re-opened only by a narrow set of later events, so a session whose activity does not produce one stays unwatched and never converges
opened: 2026-08-17
last_checked: 2026-08-17
---
# A lost inbox watch is never retried
While an identity is connected, every inbox it may read is watched, and a deposit is applied as it lands. When opening one of those watches **fails** — a broker hiccup at sign-in is enough — the failure is reported and the watch is dropped. It is then re-opened only by a later event, and the list of events that qualify is narrower than it looks:
- creating a document,
- opening an inbox on one,
- a `ReadCap` arriving through an inbox **still** being watched,
- the first read of a document in a public store.
**An ordinary write does not qualify.** A `sparqlUpdate` on a document the identity already owns fires none of these. So it is not only a read-only session that stays blind — it is any session whose activity stays inside what it already holds.
There is deliberately no polling: the deferred inbox processing states that as a design choice, and a timer here would be the wrong repair. What is missing is a recovery signal that does not depend on the identity performing a capability-bearing act.
**The visible consequence.** A document is shared with the affected identity while it sits connected, and nothing arrives. No error is raised, and the application sees an empty result rather than a failure — the same shape as having been given nothing at all.
**Workaround today:** a fresh `ensureIdentity()` re-enters and re-establishes the watches.
**Verifying it:** drive an identity to a state where one inbox watch fails to open, let the broker recover, deposit for that identity, and observe that nothing converges while it stays connected. A reproduction existed as a scratchpad probe during the work that found this and did not survive the session; rebuilding it from this description is the honest cost of that.
Predates the continuous-observation work — the earlier behaviour processed inboxes only at connection, so the same hiccup produced the same silence by a shorter route.
@@ -0,0 +1,21 @@
---
type: bug
severity: normal
summary: When the account record cannot be resolved, the inbox enumeration reports a COMPLETE list holding zero inboxes, so a total failure is indistinguishable from an identity that legitimately has none
opened: 2026-08-17
last_checked: 2026-08-17
---
# An empty inbox list can mean a failure
Enumerating the inboxes an identity may read reports, alongside the list, whether that list is **complete** or **short** — and a shortfall is logged so a lost watch is at least visible.
That signal has a hole. When resolving the account record yields nothing rather than raising, the enumeration reports a list that is **complete and empty**. Every inbox is missing, including the identity's own, and nothing distinguishes it from an identity that genuinely has none yet — which is a real and ordinary state, since a first-visit identity connects with no account.
So the worst outcome — watching nothing at all — is reported as the most benign one. This is the family this package has closed repeatedly: **a failure wearing the shape of an absence.**
The fix is not to make the enumeration throw. Reaching the registers is infrastructure and may fail; what is missing is that *resolving nothing* and *there being nothing* are the same answer here, and they are not the same event.
**Verifying it:** make the account resolution return nothing (rather than reject) and observe that the enumeration reports a complete list with no entries, and that no shortfall is logged.
Related, and distinct: [[bug_a-lost-inbox-watch-is-never-retried]] — that one is about a watch that failed to open; this one is about never learning there was a watch to open.