Repository navigation
Commit 5798baa
test(table_buffering): pin the empty-input lifecycle the extension now drives
vgi 63eb257 makes the DuckDB extension run init, combine with an EMPTY
state_ids list, then finalize when a table-buffering input is empty at
runtime. vgi-python needed no change for it: table_buffering_combine
passes state_ids straight through, finalize needs no sink state, and the
sum_all_columns fixture already writes its zero row from combine.
table_buffering_empty_input.test passes on the launch and HTTP lanes
against a unittest built from vgi 6033a63, and table_in_out/* passes on
the launch lane. vgi-python's CI fails that file only because it runs vgi
main's tests against the published extension, which predates 63eb257.
These tests drive the same sequence through the Python client with zero
process calls: sum_all_columns answers 0 0, buffer_input returns no rows,
and a combine() error still surfaces.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>1 parent 49ff227 commit 5798baa
1 file changed
Lines changed: 53 additions & 0 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
| 1 | + | |
| 2 | + | |
| 3 | + | |
| 4 | + | |
| 5 | + | |
| 6 | + | |
| 7 | + | |
| 8 | + | |
| 9 | + | |
| 10 | + | |
| 11 | + | |
| 12 | + | |
| 13 | + | |
| 14 | + | |
| 15 | + | |
| 16 | + | |
| 17 | + | |
| 18 | + | |
| 19 | + | |
| 20 | + | |
| 21 | + | |
| 22 | + | |
| 23 | + | |
| 24 | + | |
| 25 | + | |
| 26 | + | |
| 27 | + | |
| 28 | + | |
| 29 | + | |
| 30 | + | |
| 31 | + | |
| 32 | + | |
| 33 | + | |
| 34 | + | |
| 35 | + | |
| 36 | + | |
| 37 | + | |
| 38 | + | |
| 39 | + | |
| 40 | + | |
| 41 | + | |
| 42 | + | |
| 43 | + | |
| 44 | + | |
| 45 | + | |
| 46 | + | |
| 47 | + | |
| 48 | + | |
| 49 | + | |
| 50 | + | |
| 51 | + | |
| 52 | + | |
| 53 | + | |
0 commit comments