Skip to content

NOISSUE - Update reference golden fingerprints from CI - #116

Merged
SammyOina merged 1 commit into
absmach:mainfrom
nyagamunene:fix-battery-charger-golden
Oct 12, 2026
Merged

SammyOina merged 1 commit into
absmach:mainfrom
nyagamunene:fix-battery-charger-golden

Conversation

@nyagamunene

@nyagamunene nyagamunene commented Oct 11, 2026 •

Copy link
Copy Markdown
Contributor

Main's shard 1 fails in golden_svg (e.g. https://github.com/absmach/synth/actions/runs/38123330349, the #111 merge). CI draws every reference schematic about 19 paths away from the committed fingerprint (ref_battery_charger: 6115 paths vs 6134), and the test stops at the first mismatch. Same failure on #102, #108, #110, #115.

Changes fixtures/kicad-reference/svg/golden.json only: all ten fingerprints copied from the CI log of run 38181427686 (main base), not regenerated locally. An intermediate commit that printed all live hashes was reverted.

@nyagamunene nyagamunene changed the title NOISSUE - Update ref_battery_charger golden fingerprint NOISSUE - Update reference golden fingerprints from CI Oct 11, 2026
CI draws every reference schematic about 19 paths away from the committed fingerprint, so golden_svg fails on the first design it compares. All ten fingerprints are copied from the CI log of run 38181427686 (main base), not regenerated locally.

Signed-off-by: nyagamunene <stevenyaga2014@gmail.com>
@nyagamunene
nyagamunene force-pushed the fix-battery-charger-golden branch from e818989 to bf21a18 Compare October 11, 2026 23:17
JeffMboya added a commit to JeffMboya/synth that referenced this pull request Oct 12, 2026
Both depend on KiCad symbol geometry, so they can only come from a run
that has those libraries. These are CI's own numbers.

All ten golden SVG fingerprints moved by exactly nineteen paths with
their text counts unchanged — a uniform shift, so not design-specific.
`main` is red on the same thing at 79daf6f and absmach#116 is the fix in
flight there; this branch inherited it and the values here are what this
branch's own run measured. If absmach#116 lands first the overlap resolves to
whichever is live then.

`env_logger_layout` lost a wire and gained the two RX labels it had been
drawn as, the mirror of the change `sensor_logger` took earlier. The
labels are inserted where the run's name-then-coordinate order puts
them, without re-sorting the rest: an earlier attempt re-sorted the
whole block and moved two CC labels, which would have encoded a
different order from the one the layouter emits.
JeffMboya added a commit to JeffMboya/synth that referenced this pull request Oct 12, 2026
…elines

`keywords_recognized` lexes a fixed source and compares the whole token
stream, so adding `match_length` to that source without adding
`KwMatchLength` to the expected list failed it. Another guard worth
having: the keyword vocabulary cannot grow silently.

The two schematic baselines are CI's own numbers from this branch's run,
because both depend on KiCad symbol geometry this machine does not have.
All ten golden fingerprints moved — `main` is red on the same drift at
79daf6f and absmach#116 is the fix in flight there — and the feather placement
moved by hundredths of a millimetre, which is the corrected oscillator
courtyard propagating to its neighbours.

Every old value in the placement snapshot was checked against the file
before being replaced. An earlier attempt at this skipped that and had
crossed two snapshot regions in the log, which would have written
numbers no run ever produced into a release baseline.
@SammyOina
SammyOina merged commit bf93943 into absmach:main Oct 12, 2026
5 checks passed
JeffMboya added a commit to JeffMboya/synth that referenced this pull request Oct 12, 2026
Both depend on KiCad symbol geometry, so they can only come from a run
that has those libraries. These are CI's own numbers.

All ten golden SVG fingerprints moved by exactly nineteen paths with
their text counts unchanged — a uniform shift, so not design-specific.
`main` is red on the same thing at 79daf6f and absmach#116 is the fix in
flight there; this branch inherited it and the values here are what this
branch's own run measured. If absmach#116 lands first the overlap resolves to
whichever is live then.

`env_logger_layout` lost a wire and gained the two RX labels it had been
drawn as, the mirror of the change `sensor_logger` took earlier. The
labels are inserted where the run's name-then-coordinate order puts
them, without re-sorting the rest: an earlier attempt re-sorted the
whole block and moved two CC labels, which would have encoded a
different order from the one the layouter emits.
JeffMboya added a commit to JeffMboya/synth that referenced this pull request Oct 12, 2026
…elines

`keywords_recognized` lexes a fixed source and compares the whole token
stream, so adding `match_length` to that source without adding
`KwMatchLength` to the expected list failed it. Another guard worth
having: the keyword vocabulary cannot grow silently.

The two schematic baselines are CI's own numbers from this branch's run,
because both depend on KiCad symbol geometry this machine does not have.
All ten golden fingerprints moved — `main` is red on the same drift at
79daf6f and absmach#116 is the fix in flight there — and the feather placement
moved by hundredths of a millimetre, which is the corrected oscillator
courtyard propagating to its neighbours.

Every old value in the placement snapshot was checked against the file
before being replaced. An earlier attempt at this skipped that and had
crossed two snapshot regions in the log, which would have written
numbers no run ever produced into a release baseline.
JeffMboya added a commit to JeffMboya/synth that referenced this pull request Oct 12, 2026
…elines

`keywords_recognized` lexes a fixed source and compares the whole token
stream, so adding `match_length` to that source without adding
`KwMatchLength` to the expected list failed it. Another guard worth
having: the keyword vocabulary cannot grow silently.

The two schematic baselines are CI's own numbers from this branch's run,
because both depend on KiCad symbol geometry this machine does not have.
All ten golden fingerprints moved — `main` is red on the same drift at
79daf6f and absmach#116 is the fix in flight there — and the feather placement
moved by hundredths of a millimetre, which is the corrected oscillator
courtyard propagating to its neighbours.

Every old value in the placement snapshot was checked against the file
before being replaced. An earlier attempt at this skipped that and had
crossed two snapshot regions in the log, which would have written
numbers no run ever produced into a release baseline.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants