fix(cli): unify gitt issues list fill_pct math across Panel and table paths

[PfanP]

Summary

gitt issues list computed Fill % two different ways for the same field in the same function:

• view.py:95 — single-issue Panel view (--id <N>):

  fill_pct = (bounty_raw / target_raw * 100) if target_raw > 0 else 0

• view.py:146 — table all-issues view:

  fill_pct = float(Decimal(bounty_val) / Decimal(target_val) * 100)

Same field, same on-chain data, two formulas. For ratios that don't round identically under both (e.g., 1/3, 5/12, 1/7), the same issue rendered marginally different Fill % values depending on whether the operator used --id or not.

Fix

Extract a shared _fill_percent(bounty, target) helper using the Decimal path (correct, deterministic — no float-binary artifacts) and route both render paths through it.

def _fill_percent(bounty: int, target: int) -> float:
    if target <= 0:
        return 0.0
    return float(Decimal(bounty) / Decimal(target) * 100)

The agreement between the two render paths is now enforced by construction, not coincidence.

Type of Change

• Bug fix
• New feature
• Refactor
• Documentation
• Other

Reproduction (before this PR)

For an issue with bounty=1, target=3:

$ gitt i list
[ ... table renders "33%" via Decimal ... ]

$ gitt i list --id <N>
[ ... Panel renders fill_pct via raw float ... ]
# Fill %: 33.3%

The two paths agreed for most ratios but diverged on values where binary float and Decimal disagree at the rounding boundary.

After

Both render paths call _fill_percent. Format precision differs by mode ({:.0f}% for table, {:.1f}% for Panel), but both formats render the same underlying value.

Why this isn't a duplicate

• [BUG] repo_weight rounded to 2 decimals discards up to 14% of the JSON precision #652 / fix: preserve repo weight precision by removing 2-decimal rounding #654 / fix: preserve repo weight precision by removing 2-decimal rounding #655 (closed) addressed precision loss in master_repositories.json ingestion (rounding to 2 decimals) — different code path, different concern.
• fix: issues list --json --id returns null instead of structured error for missing issues #335 (MERGED) changed the not-found JSON shape on this same path; it didn't touch the math at lines 95 or 146. Existing tests from fix: issues list --json --id returns null instead of structured error for missing issues #335 are kept passing.
• No filed item touches the Fill % formula or the Panel-vs-table inconsistency.

Tests

tests/cli/test_issues_list_fill_pct.py — 13 new cases:

Group	Asserts
Unit (5)	_fill_percent(0, 0) == 0, negative target == 0, (100, 100) == 100, over-full > 100, parametric ratios (1/3, 5/12, 1/7, 2/11) match expected to 1e-12
Behavioral (4)	Table mode shows 33% for 1/3, Panel mode shows 33.3% for 1/3, both modes show 33 for the same data, JSON mode passes bounty_amount/target_bounty through unchanged

$ pytest tests/cli/test_issues_list_fill_pct.py -v
13 passed in 0.28s

$ pytest tests/cli/test_issues_list_json.py -v   # PR #335 regression tests
2 passed in 0.16s

Scope

• 2 files changed: view.py (+12 helper, 2 call-site swaps), one new test file.
• No API changes, no schema changes, no JSON contract changes.
• Single function-internal extraction; matches the pattern from PR Extract require_valid_issue_id and require_valid_ss58 CLI helpers #498 (extract validate_issue_id / validate_ss58) and PR refactor: extract duplicate network/contract resolution to _resolve_contract_and_network helper #442 (extract _resolve_contract_and_network).

Coordination

view.py:issues_list is currently being touched by #855 and #856 (both validate --id at parse time). This fix lives in the function body (the math, not the parse gate); whichever lands first, the other rebases trivially.

Related

• fix: issues list --json --id returns null instead of structured error for missing issues #335 (MERGED) — structured not_found JSON; not regressed (existing test still passes).
• Extract require_valid_issue_id and require_valid_ss58 CLI helpers #498 (MERGED) — pattern of extracting validators/helpers; mirrors that shape here.
• fix(cli): validate --id range on gitt i list to match sibling commands #855 / fix(cli): validate --id on gitt issues list (#854) #856 (OPEN) — adjacent in same function; no overlap.

Labels

bug

Fixes #896

[anderdc]

The bug premise doesn't hold — int / int * 100 and float(Decimal(int) / Decimal(int) * 100) collapse to the same float, so this isn't fixing a divergence. The extraction is a fine refactor on its own though, so I'll take it.

Scope down: keep the _fill_percent helper and the two call-site swaps in gittensor/cli/issue_commands/view.py. Drop tests/cli/test_issues_list_fill_pct.py entirely — the unit tests assert Python arithmetic, the "behavioral" tests pin format-string output, and the cross-mode test admits in its own docstring that both formulas yield the same float.

[mkdev5]

The implementation is small, but the regression test should use a value that actually fails before this change.

For bounty=1,target=3, the reported 33% vs 33.3% difference is from the existing table/panel format specs ({:.0f} vs {:.1f}), not from the math path, so these assertions do not prove the call sites were unified. Please switch the fixture to a rounding-boundary case that visibly differs under the old panel formula, e.g. bounty=23,target=80: old panel math renders 28.7%, while the Decimal helper renders 28.8%. That would give maintainers a concrete pre-fix failure and make the PR rationale much stronger.

[anderdc]

Drop tests/cli/test_issues_list_fill_pct.py; keep only the helper + two call-site swaps. The 23/80 rounding-boundary example is real, but the cosmetic impact doesn't warrant 176 lines of tests. Same scope-down as the prior review.