Uh oh!
There was an error while loading. Please reload this page.
Conversation
fryanpan added a commit
that referenced
this pull request
Jul 30, 2026
Decision input for the release conversation: what a user actually experiences, the mechanism with file:line, the device evidence, severity and frequency, and the options with their tradeoffs. Tasks #87-#91. None is fixed. None of the five is in the README's "Known limitations (v1)" list, and that is the point: the documented limitations are boundaries the design chose, while these are cases where the feature behaves worse than its own design intends. The never-stale invariant holds in all five - the gap is reliability and honesty of feedback, not speed or correctness. Ship-blocking for v1: the install-confirm fast-fail (180 s of silence ending in a message that says the user failed to confirm something never shown to them), launch-and-retry after a rebaseline reinstall (deterministic today; the fast path is dead until the user guesses), and the daemon-death reclassification (a one-line body edit becomes a ~200 s rebaseline in any Room/KSP project, 3/3 on the A56). Three corrections to the framing this doc started from, all substantiated: - shrinkDaemonForMemory() does not no-op during a mid-start respawn. Its guard is on Building, so in Degraded it runs, and its daemonEpoch++ is a CAUSE of the respawn being discarded. A memory trim during a respawn can strand the session permanently with only a log.info. - "Android never delivers PENDING_USER_ACTION to a backgrounded app" is wrong, and five places in the tree assert it. The broadcast is explicit to our own manifest receiver and is delivered (deferred at worst - our own run-2 findings say so). What is lost is the lifecycle-bound EventBus subscriber that owns the dialog, plus the Android 10+ background startActivity block. Fixing the platform theory points at the wrong repair. - The Degraded session is not entirely silent - the toolbar icon does go to the ATTENTION shape and colour. What is missing is any text, and any feedback that the tap was swallowed. Defect 3 has no device reproduction; the doc says so rather than implying one. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MkdFG7t9oNJtiayQzUQf89
fryanpan added a commit
that referenced
this pull request
Jul 30, 2026
reliability-gaps: summary cut to three bullets, table promoted to the decision content. #87 demoted - it only fires in projects with an annotation processor, which is a minority and an area Quick Build does not really support yet. #88 raised to the top: deterministic on every rebaseline reinstall, so it reaches everyone. #90 kept as blocking - a rebaseline runs long enough that backgrounding is the normal thing to do, not an edge case. Blocks-v1 is now TBD on every row for the team to decide; priority is the sort order and is stated once, not encoded three ways. Detail sections keyed by task number now that the # column is gone. Frequency cells say which rows are code-reading-only. on-device-storage-performance: rewritten as a fileable ticket and cut from 1476 to ~965 words. Kept what was run and what is slow; dropped the mechanism digression and the duplicated scope prose. perf-roadmap: provenance section cut to one line. Added a bulleted explanation of the five warm-edit stages, from the daemon source, plus two caveats the table was hiding - policy+walks sums a host-side span with a daemon-side one, and the walk time is already inside the compile RPC, so the column is not additive with total. benchmarking: dropped the "which number to quote" framing throughout. The team wants all the readings and the tradeoffs between them, not one blessed figure. Co-Authored-By: Claude <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01MkdFG7t9oNJtiayQzUQf89
fryanpan added a commit
that referenced
this pull request
Aug 27, 2026
The page proposed that #87, #89, #91 and the relink-crash gap go to v1.1, then left the table's "Blocks v1?" column reading TBD on all four rows. A proposal in a title and a TBD in a table say different things to a reader, and CodeRabbit flagged the pair as an internal inconsistency. Decision confirmed 2026-08-25: none of the four block v1. The four cells now read "No - v1.1", the title states the answer rather than asking it, and the prose no longer describes itself as awaiting confirmation. No change to any gap's evidence, root cause, or fix. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01FstXxJ5cwWPcvmhZ9vJgJ7
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
remove armeabi from lib_sources and make unknown architecture v7