Repository navigation
Conversation
`n = c[$1] * 10`, `t += c[$1]`, `print c[$1] * 2, n` declined, and `t += c[$1]` was a parse error. In the mixed walker an element read whose parent is binary arithmetic -- or the right side of a scalar `+=` / `-=` -- is now read before the action and replaced by an ssa(Ref) leaf (plawk_arith_elem_reads/12, reusing the condition reader: counter tables only, key interned as the writes intern it, absent -> 0). The rewritten action lowers through the existing emitters; admission takes the element and its ssa leaf as operands of scalar update expressions and print arithmetic. A bare copy (`n = c[$1]`) still declines (the element may be unset, spec section 3). A scalar used as an element KEY is never an unsafe strnum read (every key path uses the slot's atom id), so plawk_strnum_term_mentions/2 skips it -- otherwise `k = $1; c[k]++; n = c[k] * 2; print k` printed 0 for k (caught against gawk). Verified: ir_corpus diff over all 2591 programs: 0 changed (purely additive). tests/test_plawk_array_reads.pl: 11 (5 new); expected outputs are gawk 5.1.0's. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01QrQistdoYMNpVtecwims6g
- `for (k in c) if (c[k] > 1) print k` -- the usual braceless spelling -- was a
parse error; it now parses to the same term as the braced form, which already
lowered. A braceless `for (k in c) printf ...` parses too (its codegen is
still open, so it declines).
- Arithmetic print fields over loop-keyed elements: `print k, c[k] * 2`,
`c[k] / 4`, `c[k] + d[k]`. The iterated table is read by its slot
(value_at), any other by the loop key (get, absent -> 0), substituted as
ssa leaves and printed the awk way (integer, or %g with fdiv).
- END arithmetic over literal-key elements: `print c["a"] + c["b"]`,
`print c[5] + c[7]`, also after a for-in in a statement list. The key bytes
go through a stack buffer stored from an inline c"..." constant (this route
has no global channel); names carry `end_` so the statement-list rename keeps
them unique. plawk_assoc_print_array/2 reports the arrays inside such a field.
- END's numeric printer is factored into plawk_numeric_print_lines/6 (names as
parameters; existing names unchanged).
Counter tables only; double tables and an END scalar `if` over an element
decline.
Verified: ir_corpus diff vs the 3b commit over all 2602 programs: 1 changed,
the intended decline -> compiled `END { print c["a"] + c["b"] }` (moved from
3b's decline list to the new tests). tests/test_plawk_array_reads.pl: 15;
expected outputs are gawk 5.1.0's.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01QrQistdoYMNpVtecwims6g
This branch has not been deployed
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.
Arrays PR 3c (
docs/design/PLAWK_ARRAY_VALUE_MODEL.md§8, item 3c).Base: the feature line, not stacked. Until #4354 (3b) merges, this PR's diff also shows 3b's commit
a95163dc3; after that it shows only 3c. Merging #4354 first is cleaner, but either order lands everything on the feature line.What (matching gawk)
END { for (k in c) if (c[k] > 1) print k }(braceless)END { for (k in c) print k, c[k] * 2 },c[k] / 4,c[k] + d[k]END { print c["a"] + c["b"] },print c[5] + c[7]END { for (k in c) print k, c[k]; print "total", c["a"] + c["b"] + c["c"] }for (k in c) if (G) print …now parses to the same term as the braced form, which already lowered. A bracelessfor (k in c) printf …now parses too, but its codegen is still open, so it declines.value_at), any other table by the loop key (get, absent → 0). The reads becomessaleaves, and the value prints the awk way: an integer, or%gwith floating-point division.c"…"constant, because this route's line stream has no global channel. The names carryend_, so the statement-list renaming keeps them unique.plawk_assoc_print_array/2now reports the arrays inside such a field.plawk_numeric_print_lines/6, which takes the names as parameters; the existing names are unchanged (corpus-verified).ifover an element decline.Verification
ir_corpus.pl dump+diffvs the 3b commit over all 2602 programs: 1 changed, the intended decline → compiledEND { print c["a"] + c["b"] }. It moved from 3b's decline list into the new tests.tests/test_plawk_array_reads.pl, now 15; all expected outputs are gawk 5.1.0's.Still open (3d)
ifover an element;if (c[k] > max) { max = c[k]; mk = k });printfbodies;&&or key comparisons;🤖 Generated with Claude Code
https://claude.ai/code/session_01QrQistdoYMNpVtecwims6g