igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity
1934 commits 357 branches 1 tag 2 GiB
Commit graph

1934 commits

This branch
This branch
All branches
Author SHA1 Message Date
igneum-labs
80f4a2cda4 fast-time crossing on the hotfix f8da7515 (13:46:42Z to 14:03:21Z, 8 October 2026; the freeze fingerprint on both binaries): the crossing green on every check that reads it; the SUMMARY's three id failures were rows for epochs 12 and 13 mined during the late joiner's wait, past the observation window, with no template row (harness scope; such rows are now listed and never counted); the late joiner from genesis synced the pre-boundary blocks and waited on its executor, the hotfix's rule for a node that has synced nothing 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 14:04:29 +00:00
igneum-labs
8d29d71fba fast-time crossing: the late joiner is a reading, not a gate (under f8da7515 a node that has synced nothing keeps the executor's ten-minute wait; the cold-restart class is the node lane's unit test) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 13:44:51 +00:00
igneum-labs
3e35e65422 fast-time crossing: a late joiner from an empty datadir after the main loop (the chain over eleven minutes old, its executor's sink at genesis: the 0.3.25 sink-age guard's shape, service.rs:1131), must reach n0's sink with no lock fault and no IBD refusal; --late-join-s caps the wait, 0 turns it off 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 13:36:58 +00:00
igneum-labs
10f2cad9d0 fast-time crossing on the 0.3.25 pair c9ad753a (12:59:04Z to 13:11:43Z, 8 October 2026; the freeze fingerprint cbc5bd0a... on both binaries read before the lease): SUMMARY PASS on every check, the gate's record on the pin 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 13:12:10 +00:00
igneum-labs
3fd20f60e8 fast-time crossing on the rebuilt 0.3.25 pair 6ccaf9e9 (12:44:45Z to 12:57:35Z, 8 October 2026; the freeze fingerprint cbc5bd0a... on both binaries read before the lease): SUMMARY PASS on every check, the gate's record on the pin 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 12:58:16 +00:00
igneum-labs
5cc9528b0b fast-time crossing on the 0.3.25 pin commit d5b68fae (12:16:13Z to 12:29:05Z, 8 October 2026): SUMMARY FAIL on v5_ids_equal_the_cli_v5_id, a node finding: on epoch 9's seed (attempt 3 under the freeze's acceptance) the pair's miner and nodes drew 65b57e3b847d362e while the freeze CLI dba17a0a and the 03e5d172 CLI draw ebf64b32e2d84c5b; the pair's igneum-pow is not the freeze's on an acceptance rule; every other reading green 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 12:32:08 +00:00
igneum-labs
8b8d8b34f5 fast-time crossing on the 0.3.25 pair 42ce0f07, the clean run (12:10:07Z to 12:22:57Z, 8 October 2026): SUMMARY PASS on every check 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 12:23:16 +00:00
igneum-labs
69b76d9aa7 fast-time crossing on the 0.3.25 pair 42ce0f07 (11:55:54Z to 12:08:50Z): every reading green (rung 1 at epoch 6, v5 at epoch 8 at rung 1, 12 of 12 ids equal to the freeze CLI's, the stale node refused, the restart step resynced in 19.1 s with the catch-up done, four sinks equal); the one false check was the harness's: the last epoch's row had one miner's line when the run ended at that epoch's boundary; that row is now read by its id alone 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 12:09:30 +00:00
igneum-labs
1b4564d6fb fast-time crossing on the 0.3.25 pair 39f127a1 (node code and object byte for byte e0644958's; 09:42:25Z to 09:54:40Z, 8 October 2026; igneum-pow at the freeze dba17a0a): SUMMARY PASS on every check, the gate's record on the pin; the proving fee ceiling's two fields absent from the 60x file, the ceiling at never in this run 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 09:55:02 +00:00
igneum-labs
7dcec20e43 fast-time crossing on the 0.3.24 morning pin 5b673577 (08:35:18Z to 08:47:45Z, 8 October 2026; igneum-pow at the freeze dba17a0a): SUMMARY PASS on every check, the gate's record on the shipped commit 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 08:48:03 +00:00
igneum-labs
4b78645751 fast-time crossing on the 0.3.24 re-cut dfbd1e10 (23:58:56Z to 00:12:57Z, 8 October 2026; igneum-pow at the freeze dba17a0a): SUMMARY PASS on every check, the gate's record on the shipped commit (rung 1 at epoch 6, class v5 by signal at byte 6 from epoch 8 at rung 1, 11 of 11 ids, the stale node refused, n2 restarted across the boundary on its own datadir and resynced in 36.2 s with the catch-up done, four sinks equal) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-08 00:13:12 +00:00
igneum-labs
0c28401425 fast-time crossing on the 0.3.24 re-cut c9e385eb (22:36:25Z to 22:49:32Z, igneum-pow at the freeze dba17a0a): SUMMARY PASS on every check, the gate's record on the shipped commit (rung 1 at epoch 6, class v5 by signal at byte 6 from epoch 8 at rung 1, 11 of 11 ids, the stale node refused, n2 restarted across the boundary on its own datadir and resynced in 28.1 s with the catch-up done, four sinks equal) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 22:49:51 +00:00
igneum-labs
648ef34b74 fast-time crossing on pair c8f9b383, the clean run (21:22:49Z to 21:36:35Z): SUMMARY PASS on every check: class v4 sub-version 3 from genesis at rung 0, rung 1 by signal from epoch 6, class v5 by signal at byte 6 counted exactly from epoch 8 at rung 1 before the floor, 11 of 11 ids equal to the freeze CLI's, the stale node refused from the first refresh, n2 restarted across the boundary on its own datadir and resynced in 12.1 s with the catch-up done, no lock fault, four sinks equal 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 21:36:53 +00:00
igneum-labs
20fb42f1a2 fast-time crossing on pair c8f9b383 (21:07:42Z to 21:21:29Z, igneum-pow at the class v5 freeze dba17a0a): the restart step holds (the catch-up anchored at the last validated honest header, done after 12 s with 18 deferred headers validated, IBD completed, n2 at n0's sink 18.1 s after its start, nothing of its own mined during the catch-up, 82 blocks after, four sinks equal at 660); 11 of 11 ids; the stale node 92 of 92 refused; two harness checks reworded (the IBD-refused regex matched the catch-up's announcement; the hold is read from what was accepted, a refusal line only when the miner found a block inside the window) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 21:22:28 +00:00
igneum-labs
2b5823aa3d fast-time crossing on pair 432ea3d6 (20:53:12Z to 21:07:32Z): the node-side hold reads (119 Reject(IsInIBD) refusals of the restarted node's own miner during the catch-up, 0 accepted after, no fork); the catch-up anchor stall as on 63524e28; the hold check measures from the catch-up's start line, not a done line it cannot have while the anchor stalls 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 21:08:31 +00:00
igneum-labs
3bc5885f35 Merge class-v5 dba17a0a (the class v5 FREEZE: the igneum-pow pair 4 c8f9b383 pairs with) into v5-fasttime 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 21:01:15 +00:00
igneum-labs
dba17a0a4f class v5 FREEZE: the suite's green and the gate on the record (section 0), the shadow rule's measured trigger reading, the next commit named 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:53:07 +00:00
igneum-labs
b999554151 fast-time crossing on pair 63524e28 with the restarted miner held (20:37:24Z to 20:51:22Z): the chain crossed as before; no fork from n2; the IBD catch-up anchored its body sync at n2's own sink (chain block 394), the executor stayed at tip 394, the relayed honest blocks were orphaned 96 times, no catch-up done line; the harness reads the node-side hold shape for pair 3 (--restart-miner-hold-s 0: the miner's submits refused during the catch-up, no block of its accepted before the catch-up done line, blocks after it), the hold refusals counted apart from PoW refusals, the node log's clock parsed 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:52:23 +00:00
igneum-labs
7965f5bafa class v5 design page: section 14's final total from adv-accept (seven of seven live hot sets refused; the three mild concentrations at about 1.0004x) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:49:23 +00:00
igneum-labs
886be84001 class v5 shadow rule test: the scan reads the drawn blocks' maximum removable share and redraws (none over the bound in 6,000 first draws under the census arm); the liveness claim is the synthetic over-bound block through the same predicate 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:45:16 +00:00
igneum-labs
1fabc0842a class v5 design page: section 14's running total (adv-accept's seed 1738 refused; six of six live hot sets under the floor) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:39:15 +00:00
igneum-labs
03c3f88046 class v5 design page: section 14's record sentence is the defender's ruling (the hot-set half closed, the residual named with prevalence, gain and lever) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:38:42 +00:00
igneum-labs
7eb3ad69ce class v5 design page: section 14 carries the five programs' ratios and the record's sentence (the floor closes the class on every measured hot set by X_f >= f and misses the shadow-block-write residual, named with its instrument and bound); the residual as a gap row 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:38:29 +00:00
igneum-labs
764a41d9e7 class v5 shadow rule test: the scan is over the candidate draw at attempt 0 (the full draw costs seconds per candidate under the 2^20 ratio pass; the first form ran over an hour on box 2) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:37:52 +00:00
igneum-labs
ceceaf64d6 class v5 harness runner: the box side in a script run by path (main's rule of 21:33 UK: no redirect-truncate inside a shell command string) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:36:11 +00:00
igneum-labs
49845723fe fast-time crossing on pair 63524e28 (20:19:59Z to 20:34:12Z): rung 1 at epoch 6, v5 at epoch 8 at rung 1, the stale node refused 132 of 132, the prelude v4; the restart step: the relay refusal gone, the IBD catch-up engaged and never completed because the node's own miner, restarted at once with unsynced mining on, extended its branch from the split and the executor followed it; the harness holds the restarted node's miner until its sink equals n0's (cap 300 s) and reads the catch-up done line 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:35:41 +00:00
igneum-labs
cb15a8d3dc fast-time, the matched pair 959b57c9 (igneumd 519ee6c4..., igneum-pow 03e5d172) under lease pool class v5: the crossing run 19:49:33Z to 20:04:00Z (the prelude reads v4 at epochs 0 and 1 on b1680b57; rung 1 by signal from epoch 6 at 19:56:38Z; class v5 by signal from epoch 8 at rung 1 at 19:58:31Z, 3 of 3, byte 6 at 9,985 bps; 11 of 11 program ids equal to the CPU verifier's; the stale node refused 122 of 122 after the first refresh; the restart step FAIL: the IBD catch-up engaged and did not complete, n2 on its own fork), the digest case PASS at 20:08:30Z (one digest c0d6998e for the v5 key absent, at never and on the 0.3.23 control 2720d8d2; the set key 8d8f6208 refused with the line in the control's log; 180 s of mixed mining, 79 new and 101 old blocks, 0 rejected, one sink at 180); the second-step check reads the rule (a step at or after the earliest allowed epoch is the rule, 12 here, decided after the window closed) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:08:49 +00:00
igneum-labs
0d5570e299 class v5 (c'''): the census number in the record (4,600 f8 seeds: 2.435 percent of sub-version 3 accepted programs under 0.995, attempts mean 2.174 to 2.248, 0 class v5 programs under the floor; 0.99 would pass the exemplar), the flip-stale PASS on the matched binaries 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 20:04:35 +00:00
igneum-labs
f8e6d9f84a class v5 harness: the stale miner's fall-off is judged by the block's DAA score (a nonce found on a pre-boundary template and submitted after it is a flip-epoch block, valid; the wall-clock count failed the 20:4x UK run on one such block), nonces matched as the nearest double the JSON RPC carries 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:51:43 +00:00
igneum-labs
9ac364b722 class v5 census runner: lease class v5 (the pool's priority classes, lease 5d84d644) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:41:44 +00:00
igneum-labs
658f6b9195 class v5 design page: section 14 carries adv-accept's multiplicity reading at the acceptance sample 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:40:48 +00:00
igneum-labs
78714e59d5 class v5 design page: section 14 carries adv-accept's class v5 check of the exemplar (the hot set does not move under the state dataset) and the known-failed test's reading 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:39:16 +00:00
igneum-labs
4d6218ee84 class v5 census runner: pick the test binary by --list (the CLI binary shares the igneum_pow-<hash> name; the first run held 64 pool cores for 9 minutes on the CLI's usage text) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:36:21 +00:00
igneum-labs
4c081ffc4d fast-time digest case: the first run (19:25:30Z: one digest c0d6998e for the v5 key absent, at never and on the pre-v5 binary; the set key at 8d8f6208 with no peer and the refusal line in the pre-v5 node's log; 0 rejected) and its one fault: the pre-v5 miner was given --exec-rpc, which it does not have, and exited on usage; the new miner alone made 19 blocks in 180 s on a box at load 360. The old miner takes no exec flag, both take --threads (default 4), a usage exit is a named check 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:26:48 +00:00
igneum-labs
7bb87b096f class v5 census runner: the test binary is the extension-less igneum_pow-<hash> (the first pick read an object file) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:26:40 +00:00
igneum-labs
7acccc31e5 class v5 census runner: the lease label without apostrophes (the remote shell read the (c''') mark as quotes) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:25:53 +00:00
igneum-labs
905f57967f fast-time crossing: the first cross run's summary (19:22:01Z: rung 1 at epoch 6, v5 at epoch 8 at rung 1, every id equal to the CPU verifier's on 13 of 13 rows, byte 6 at 9,985 bps; the restart step's IBD refusal and fork) and five harness-scope corrections it showed: a fork row (a seed that is not n0's template seed) is listed and never counted; the step line is logged twice per step on the 0.3.23 line, so a second step is a new (epoch, rung) pair; a submit RPC timeout is counted apart from a PoW refusal; the stale node's accepted block on an old template after the refresh is the previous epoch's; the miners expected on a row allow for the restarted node 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:24:24 +00:00
igneum-labs
15eeb5b5f6 class v5: the (c''') census runner on box 2 through the pool lease (the suite's own test binary, the log copied under the harness dir) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:24:03 +00:00
igneum-labs
1a627e85af fast-time crossing: the known-failed case's FAIL (19:04:01Z, 22 checks named) and the first cross run's evidence (the chain crossed: rung 1 by signal at epoch 6, class v5 by signal at epoch 8 at rung 1; the restart step FAILED: n2 restarted on its own datadir across the boundary could not IBD, 'proof-of-work check refused, class v5 needs the execution state after the epoch's seed block', 13 attempts, its own fork; the end-of-run reads hung on its exec RPC); the harness: a 5 s timeout on every stream fetch, the IBD refusal after the restart as a named check, the end-of-run reads under a 180 s cap 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:22:16 +00:00
igneum-labs
aa82ffefb0 fast-time: the crossing harness asks the CLI for the row's own class (the v3 prelude rows printed MISMATCH against a v4 id; no check read them) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:14:56 +00:00
igneum-labs
b86ca261c4 Merge build/class-v5 03e5d172 into v5-fasttime (igneum-pow at the (c''') acceptance commit, the pair for the rerun on fork b1680b57) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:14:56 +00:00
igneum-labs
9705a022ef class v5 harness: the miners stop before the final sink and count read (a node's own newest block is a tip its peers have not seen yet; the 19:50 UK run compared sinks mid-propagation) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:10:20 +00:00
igneum-labs
d0939373b0 class v5 design page: section 14 (the acceptance rule's class v5 part on the residual hot-set class, the two fixes, the floor taken), the witness row done, the class-walk exception, the harness skew and override-field rows 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:07:32 +00:00
igneum-labs
03e5d172ad class v5 acceptance (c'''): the per-site distinct-index floor raised to 0.995 on the ratio pass's own run, keyed on the state flag; known-failed first on adv-accept's seed 100767 (program 9d68e6286fc817d4, site 6 at 0.9919), the census test behind the floor 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 19:05:48 +00:00
igneum-labs
90121a4042 fast-time: lib/override.mjs (the override as text: duplicate keys collapsed, keys dropped by line, the binary probed for the keys it does not know and those dropped, one prune set for every file of a case); the crossing and digest harnesses take it (the first run at 18:47Z: fork 546fe4b5, on release-0.3.23-node 43360992, refuses the 60x file's sig_scheme and three later keys that master's newest fork line added; the digest harness no longer imports the crossing harness, which would have run it) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 18:49:32 +00:00
igneum-labs
6c3e2fd2ea class v5 harness: drop the override fields the fork binary does not know, named in the log (master's 60x file carries every lane's fields; OverrideParams is deny_unknown_fields; the binary's field set is its source's) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 18:49:27 +00:00
igneum-labs
013b8da18b class v5 shadow rule test: the synthetic rotate cases follow the census arm (two rotl by one amount merge, a rotl and a rotr do not, two rotr by one register merge) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 18:44:51 +00:00
igneum-labs
c569ed0783 class v5 harness: the accepted-after parser reads the miner's epoch-second line clock (it matched HH:MM:SS and counted nothing: the v5-fasttime lane's finding, 20:3x UK), with its own known-finished check (the honest miners must show accepted lines after the flip); the page's PASS rows say which two counts were parse misses and what they rest on until the re-run 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 18:38:42 +00:00
igneum-labs
f6f9719a79 class v5 shadow rule: the rotate-merge arm is the census's (the same rotate with the same amount or amount register, the source untouched), not every rotate pair, since every op reads its own destination; the tests allow what acceptance does (it reads the shadow, so a redrawn block can move the accepted attempt and the base program): an unredrawn seed is the amended v4 draw for draw, a redrawn seed's first v4 block was over the bound, and the redraw rate stays near the census's 4e-3 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 18:37:58 +00:00
igneum-labs
41722e4c19 fast-time: the ladder and class v5 together, the Devnet 3 crossing rehearsal (class-v5-crossing.mjs: class v4 sub-version 3 from genesis at rung 0, one rung by miner signal over a 45-DAA ladder window at epoch 6, v5 by signal at byte 6 counted exactly over a 60-DAA class window at epoch 8 before the floor, the stale node on --freeze-state refused from the first refresh, n2 restarted on its own datadir across the boundary, one program id per epoch against the CPU verifier with the window's stream fetched during the run, the Devnet 3 genesis id e5a4ac5978462156 as the CLI's self-check); class-v5-digest.mjs (the v5 field absent, at never and set: one digest for the first two on the new and the pre-v5 binary, the set field refused at the handshake with the line in the pre-v5 node's log, mixed mining with 0 rejected); override-60x.json: the six top-level keys the 18:53 merge doubled collapsed to one copy (the node refuses a duplicate key; tools/ci/override-json-check.sh green again) 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
 2026-10-07 18:37:12 +00:00