I implemented Panini's order-independence principle from…
1 reactions
I implemented Panini's order-independence principle from
the Ashtadhyayi as a programming language — same source
compiles to 7 targets with an invariant semantic fingerprint
Panini's Ashtadhyayi (5th c. BCE) has a property that
modern PLs mostly lack: order-minimization. The same
grammatical derivation is reachable through multiple
rule-application paths. The grammar is a system, not
a sequence.
I built Sadhana around that principle. A .sadhana source
file declares entities and relations in any order — the
compiler builds a hypergraph, computes a Canonical Meaning
Kernel (CMK), and generates code in any of 7 target
languages. The CMK is identical regardless of declaration
order and regardless of which backend you target.
The CMK is a five-part hash: entity topology, relation
patterns, constraint predicates, semantic role signature,
and a synthesis hash of all four. It functions as a
formal proof of semantic equivalence across representations.
Also implemented: a reversible encoding called Bija
(Sanskrit: seed) that compresses the meaning graph to
\~10:1 and decodes back completely — different from a
hash in that it's invertible.
v1.0 is published with a DOI:
https://doi.org/10.5281/zenodo.18846465
GitHub: https://github.com/nickzq7/Sadhana-Programming-Language
#technology
the Ashtadhyayi as a programming language — same source
compiles to 7 targets with an invariant semantic fingerprint
Panini's Ashtadhyayi (5th c. BCE) has a property that
modern PLs mostly lack: order-minimization. The same
grammatical derivation is reachable through multiple
rule-application paths. The grammar is a system, not
a sequence.
I built Sadhana around that principle. A .sadhana source
file declares entities and relations in any order — the
compiler builds a hypergraph, computes a Canonical Meaning
Kernel (CMK), and generates code in any of 7 target
languages. The CMK is identical regardless of declaration
order and regardless of which backend you target.
The CMK is a five-part hash: entity topology, relation
patterns, constraint predicates, semantic role signature,
and a synthesis hash of all four. It functions as a
formal proof of semantic equivalence across representations.
Also implemented: a reversible encoding called Bija
(Sanskrit: seed) that compresses the meaning graph to
\~10:1 and decodes back completely — different from a
hash in that it's invertible.
v1.0 is published with a DOI:
https://doi.org/10.5281/zenodo.18846465
GitHub: https://github.com/nickzq7/Sadhana-Programming-Language
#technology
1
| 2,000 mlx | |
| 3 | |
| 1 |