back to table · edit · history · where entries came from · files · download
1516 bytes, as of the version from 2026-10-26 04:16 (current). Recorded here, not run.
# Build-environment observations, T505
The isolated Sage runner used Python 3.12.3, Sage 10.9 and numberdb 0.1.10.
The checkout's private `_write.provenance_headers` helper was not present in
that installed client. Importing it stopped the first fill wrapper before
any credential read or API write. The wrapper was changed to send explicit
provenance headers using the existing Client.submit method. The published
reproduction scripts do not depend on that helper. This is a deployment
version difference, not a proposed change to the mathematical table skill.
The API's final document write and all attachments report the same document
digest. Attachment writes use the same run identifier and the responses report
amendment after the first attached file. The numerical entries were sent once.
The attachment list links to HTML viewers (`/files/T505/checks.py`). Raw bytes
are returned by adding `?raw=1`. The first checksum compared viewer HTML with
source; all source and evidence checksums matched after using the raw links.
`queue.py built 209 ... T505` ticked the correct theta-four checklist line,
but its request-matching helper also treated theta-three as the same subject
when gathering answered requests. It posted comments on #206 and #208, both
already closed. Those new comments were corrected to identify T505 as an
unpublished theta-four draft and say explicitly that theta-two is still
outstanding in #209. No issue state was changed from its pre-run state.
The pipeline source was not edited.