Repository navigation
Conversation
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This construction produces the current native flat guide from physical binary descriptors, feeds the actual guide to literal INIT and CLEAR, and retires it completely. The fixed seven-tape lifecycle costs at most 72L instructions and retains at most 7L visited cells on its stated fresh windows, where
L=N(2W+3). Prior reserved spans remain charged separately. No new kappa, complete recursive allocator or framework discharge is claimed.Read the all-size proof, complete emitted tables, checked receipt, emitter, two-interpreter verifier, retained inputs and copied source pins. This supplies the guide-production boundary left by PR #7666 for the unchanged format and reset graphs in PR #7663, exact source
b81f40997de646de30749d367484d89e59fc3013.The caller supplies positive canonical binary record count N and signed component width W. Each record contains W real bits, a separator, W imaginary bits, a separator and one represented blank. The emitted producer preserves both descriptors, writes every guide cell, and erases its binary/unary work and installed markers. The actual guide then controls both reset programs. Retirement erases its right marker, complete body and left marker while returning the head. All heads return, and borrowed exterior cells remain unchanged. Interior physical BLANK is a guide value, not a terminator.
The producer has 210 instructions, each reset has 40, and retirement has 16, independent of N and W. Runtime instructions see only scanned tape symbols. They use no decoded host dimensions, callbacks or variable-distance moves. The retained binary counter proof gives
c(x)=24x-7 popcount(x)+15 bitlength(x)+12. The complete lifecycle has exact costc(N)+c(W)+72NW+111N+7W+68. Its exact visited capacity is2L+N+W+bitlength(N)+bitlength(W)+max(bitlength(N),bitlength(W))+14. Erased backing remains included; previously reserved spans use the maximum of old and current capacity on each tape.Use Python 3.12 and the standard library:
Expected output includes:
Normal and optimized isolated replay passed with identical tables. All 42 shapes agree with a separate dense opcode interpreter on complete cells, heads and instruction counts. They cover binary shrink boundaries, widths 1,2,3,17 and full views. The N=512,W=1 lifecycle takes exactly 106258 instructions and retains 5668 cells. Six controls detect wrong zero output, skipped represented-blank writes on dirty guide cells, a premature boundary, a missing return, incomplete guide retirement and a noncanonical binary descriptor. Committed bytes, source ancestry, official export and read-only Git-free optimized replay passed. Hosted CI did not certify this package. The plan job failed in the shared
test_current_ci_section_matches_selected_record: the base guide names frontier6030 while the selected record requiresmake frontier7000-verify. The log reports 760 tests, one failure and 24 skipped; package replay was skipped. This package changes no shared CI file. Local replay remains the scientific evidence.Limits: the three conversion work tapes start blank. Guide and target have exact granted windows; arbitrary old private tails require separate handling. Root production of N and W, numerical work between resets, native adapters, the actual recursive owner map, cross-depth auxiliary stock and the whole physical tariff remain open. The source's full application and selected multiplier were not replayed. Failure need not restore work or heads. All five inherited assumptions and the selected frontier remain unchanged.
Base:
e54523517d0f03e58352e647c403a5af3e525a57. David Arias supplied the unchanged machine, assembler and counter, as credited in the retained notice and source headers. @thu-nmrc supplied the PR #6719 package that retains them; its delayed-cover root theorem has a different consumer and is not repeated here. @aliromman supplied the PR7663 guide/reset source with substantial OpenAI Codex assistance. @CubeJerry supplied its selected primitive and prior typed reset in PR #7586. The copied notice preserves the native lineage. Ali Romman directed this new emitted lifecycle, proof and verification with substantial OpenAI Codex assistance. New work is Apache-2.0; predecessor sources are unchanged.