create-majlis 0.8.1 → 0.8.3
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/index.js +5 -5
- package/package.json +1 -1
package/dist/index.js
CHANGED
|
@@ -99,7 +99,7 @@ Read source code at the specific locations relevant to your change. Do NOT
|
|
|
99
99
|
read the entire codebase or run diagnostic Python scripts. If the synthesis
|
|
100
100
|
says "lines 1921-22" then read those lines and their context. That's it.
|
|
101
101
|
|
|
102
|
-
Do NOT read raw data files (fixtures/, ground truth
|
|
102
|
+
Do NOT read raw data files (fixtures/, ground truth, test data). The synthesis
|
|
103
103
|
has the relevant facts. Reading raw data wastes turns re-deriving what the
|
|
104
104
|
doubt/challenge/verify cycle already established.
|
|
105
105
|
|
|
@@ -138,7 +138,7 @@ Do NOT iterate. Do NOT "try one more thing." The adversary, critic, and verifier
|
|
|
138
138
|
exist to diagnose what went wrong. The cycle comes back to you with their insights.
|
|
139
139
|
|
|
140
140
|
## Off-limits (DO NOT modify)
|
|
141
|
-
- \`fixtures/\` \u2014 test data, ground truth,
|
|
141
|
+
- \`fixtures/\` \u2014 test data, ground truth, reference outputs. Read-only.
|
|
142
142
|
- \`scripts/benchmark.py\` \u2014 the measurement tool. Never change how you're measured.
|
|
143
143
|
- \`.majlis/\` \u2014 framework config. Not your concern.
|
|
144
144
|
|
|
@@ -721,9 +721,9 @@ falsifiable structural constraint that blocks future experiments from repeating
|
|
|
721
721
|
## Constraint Quality
|
|
722
722
|
|
|
723
723
|
Good constraints are specific and block future repetition:
|
|
724
|
-
- "
|
|
725
|
-
- "Relaxing
|
|
726
|
-
- "
|
|
724
|
+
- "Binary search fallback in sortedMerge() is O(n log n) worst-case when input contains >40% duplicates \u2014 measured via 13-run sweep, ceiling 0.67x baseline"
|
|
725
|
+
- "Relaxing the pruning threshold in recursiveSplit() causes false positives on sparse inputs (accuracy 95->72.5%)"
|
|
726
|
+
- "Cyclic dependency in the module graph prevents topological sort \u2014 requires at least one back-edge cut"
|
|
727
727
|
|
|
728
728
|
Bad constraints are vague and useless:
|
|
729
729
|
- "Didn't work"
|