Extensive changes from yours - I ran the full test suite about five or six times and only got the failures three or so times - I'm going to try to look at fixing it right - I think the second is mostly harmless? Probably has to do with BTC/PKT differences. The new tests
that I've added are working and passing, so that's a good sign, at least. I have more to add as well.
Oddly, the MORE threads and CPU's I tell it to run with, the less it seems to produce the failures. Grrr....
Still left to merge is my total revamp of GoLevelDB (with the caching working and all race conditions fixes, and it's been exercised with TB's and TB's of data without issue - so not just "all the races fixed"), but also, the all new testing framework, which includes leak detection for all tests, the new BBoltDB optimizations, etc. I still have about 4 or 5 PR's to merge with that branch. I'll also look at how difficult it would be to merge the rest of my tree local tree which has some additions ported from
BitCoin Core and Bitcoin Cash, as well as my initial implementation of the UDP fastblock transfer mode, and some other notable performance improvements and changes, but the remainder are mostly backports from mainline and decred.
I also have adjustments to the INV and ban system, which fixes that too early triggered "misbehaving peer" warning (just need to break out), and the Mage build is just about ready to merge in as well.
~I hope to get it fully in-shape tomorrow, but, feel free to poke around if you like.~
(~FYI,~ PR's left to review are and decide if I want to merge are #111, #229, #231, #232, #234, #235) ~, and the notes in #251 - everything is in my local tree.)~
Even better, I promise I'll have the automatic package builds working for macOS, Linux, and Windows ~tomorrow~ today as well as the coverage reporting analysis.
johnsonjh
commented
3 years ago
@cjdelisle
[via GitLab] - From #254:
It's up now, the develop-johnsonjn branch, also, the same is available over at Gridfinity GitLab.
I'm still getting random failures on these two tests - I'll look into fixing it tomorrow, properly:
---
FAIL: TestLightningWallet/btcwallet/btcd:signed_with_tweaked_pubkeys---
FAIL: TestWeightEstimateExtensive changes from yours - I ran the full test suite about five or six times and only got the failures three or so times - I'm going to try to look at fixing it right - I think the second is mostly harmless? Probably has to do with BTC/PKT differences. The new tests that I've added are working and passing, so that's a good sign, at least. I have more to add as well.
Oddly, the MORE threads and CPU's I tell it to run with, the less it seems to produce the failures. Grrr....
Still left to merge is my total revamp of GoLevelDB (with the caching working and all race conditions fixes, and it's been exercised with TB's and TB's of data without issue - so not just "all the races fixed"), but also, the all new testing framework, which includes leak detection for all tests, the new BBoltDB optimizations, etc. I still have about 4 or 5 PR's to merge with that branch. I'll also look at how difficult it would be to merge the rest of my tree local tree which has some additions ported from BitCoin Core and Bitcoin Cash, as well as my initial implementation of the UDP fastblock transfer mode, and some other notable performance improvements and changes, but the remainder are mostly backports from mainline and decred.
I also have adjustments to the INV and ban system, which fixes that too early triggered "misbehaving peer" warning (just need to break out), and the Mage build is just about ready to merge in as well.
~I hope to get it fully in-shape tomorrow, but, feel free to poke around if you like.~
(~FYI,~ PR's left to review are and decide if I want to merge are #111, #229, #231, #232, #234, #235) ~, and the notes in #251 - everything is in my local tree.)~
Even better, I promise I'll have the automatic package builds working for macOS, Linux, and Windows ~tomorrow~ today as well as the coverage reporting analysis.