docs(brain): refresh Current state; sharpen AG tasks (D cut to one question, add G/H) - #99
Merged
Merged
Conversation
Two inconsistencies found while re-reading the doc: - "Current state" still said the grid was on 0.11.0 awaiting redeploy, and did not mention the biggest behaviour change since: remote preprocessing (#42) is now ON BY DEFAULT. Anyone reading only that section got the wrong picture of how the client behaves. Now states it explicitly, with the measured wins (-42% client CPU, 3.19x wall-clock on 101 TUs), the automatic fallbacks, the SUCO_REMOTE_PREPROCESS=0 escape hatch, and the caveat that it is NOT in the 0.12.0 binaries. - Open items claimed #43 "still needs the worker-side wiring" — that landed in #55, as the v1.0.0 section 15 lines above correctly says. Narrowed to what is actually left: per-job /tmp isolation. Same standard the AG status record is held to: a doc that contradicts itself is worse than one that admits a gap. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Asking the full Task D twice hasn't produced the Windows-worker evidence, so a third identical ask is unlikely to. Reduced to one unambiguous question: toggle WIN-DEV in, compile one TU, paste the WIN-DEV log lines — does it say "Compiling direct RPP job" or not. Yes / No / Can't-run are ALL stated as acceptable answers; the only wrong outcome is a green tick without the log line. Task F reduced the same way (has SUCOGrid.dll ever loaded in a running VS2022 — yes or no). Adds two tasks that match AG's demonstrated strengths instead of leaving it stuck: - Task G: installer end-to-end with the NEW default (V3 is on by default, so an installed client behaves differently from the 0.12.0 binaries — validate the artefact users actually get; AG did this well for 0.11.0). - Task H: re-run the 101-TU benchmark 3x per mode and report the median, so the 3.19x figure doesn't rest on a single run before it lands on a README. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
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 free
to 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.
Two commits that were waiting on connectivity.
1. Current state was stale/contradictory — it still said the grid was on 0.11.0 awaiting redeploy and never mentioned the biggest behaviour change since: remote preprocessing is ON by default. Now stated with the measured wins (−42 % client CPU, 3.19× wall-clock), the automatic fallbacks, the
SUCO_REMOTE_PREPROCESS=0escape hatch, and the caveat that it is not in the 0.12.0 binaries. Also fixed an Open-items entry claiming #43 "still needs the worker-side wiring" — that landed in #55; narrowed to what's actually left (per-job/tmpisolation).2. AG tasks sharpened. Asking the full Task D twice hasn't produced the Windows-worker evidence, so it's cut to one unambiguous question: toggle WIN-DEV in, compile one TU, paste the WIN-DEV log — does it say
Compiling direct RPP job? Yes / No / Can't-run are all stated as acceptable; the only wrong outcome is a ✅ without the log line. Task F reduced the same way. Plus two new tasks matching AG's demonstrated strengths: G installer end-to-end with the new default (an installed client now behaves differently from the 0.12.0 binaries), and H re-run the 101-TU benchmark 3× per mode and report the median before 3.19× becomes a public claim.Doc-only.
🤖 Generated with Claude Code