You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Line 1 is the opening ---, which is the most basic construct the format has,
so this does not look like an edge case in the grammar.
Source: Attacca at website/src/pages/index.astro.
2. The warning misdirects every reader
extract.py:5636 (0.9.44):
f"may be partially extracted: {_shown}{_more} (#2551)",
(#2551) is a literal in the shared message, emitted regardless of language.
The surrounding comments already know this — line 5619 refers to "the genuine #2551 Kotlin one-line-body" case — but the reference is still attached to all
of them.
The cost is concrete: the citation is the only lead the message offers, and it
sends the reader to a closed Kotlin issue. We initially recorded this astro
failure as "already tracked upstream" on the strength of that number, and only
found otherwise by opening it.
Suggestion: drop the hardcoded number, or make it language-aware, or point at a
stable docs page about partial extraction instead of a single issue.
Why partial extraction deserves a loud signal
"May be partially extracted" means an unknown number of nodes are missing from
the graph, and the run still exits 0. For a knowledge graph that is silent data
loss, and a consumer cannot tell a sparse file from a truncated one. A count of
what was dropped, even approximate, would make it actionable.
Summary
Two related problems, observed on 0.9.44.
---frontmatterblock, so the file is only partially extracted.
(#2551)for every language, andKotlin: bundled grammar rejects one-line type bodies (class C { val v = 1 }) — 21/541 files partially extracted, exit 0, no warning #2551 is a closed, Kotlin-specific issue. Anyone who follows the pointer
lands on a resolved problem that has nothing to do with their file.
1. Astro frontmatter fails at line 1
The file opens with the standard Astro component-script fence:
Line 1 is the opening
---, which is the most basic construct the format has,so this does not look like an edge case in the grammar.
Source:
Attaccaatwebsite/src/pages/index.astro.2. The warning misdirects every reader
extract.py:5636(0.9.44):f"may be partially extracted: {_shown}{_more} (#2551)",(#2551)is a literal in the shared message, emitted regardless of language.The surrounding comments already know this — line 5619 refers to "the genuine
#2551 Kotlin one-line-body" case — but the reference is still attached to all
of them.
The cost is concrete: the citation is the only lead the message offers, and it
sends the reader to a closed Kotlin issue. We initially recorded this astro
failure as "already tracked upstream" on the strength of that number, and only
found otherwise by opening it.
Suggestion: drop the hardcoded number, or make it language-aware, or point at a
stable docs page about partial extraction instead of a single issue.
Why partial extraction deserves a loud signal
"May be partially extracted" means an unknown number of nodes are missing from
the graph, and the run still exits 0. For a knowledge graph that is silent data
loss, and a consumer cannot tell a sparse file from a truncated one. A count of
what was dropped, even approximate, would make it actionable.