Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 19:01

    quentinlesceller on master

    Remove usage of try! macro, use… (compare)

  • 19:01
    quentinlesceller closed #3097
  • 16:00
    hashmap opened #3097
  • 16:00
    hashmap labeled #3097
  • 09:27
    antiochp labeled #3096
  • Oct 20 18:34
    JosephGoulden synchronize #3096
  • Oct 20 18:11
    JosephGoulden synchronize #3096
  • Oct 20 18:01
    JosephGoulden opened #3096
  • Oct 19 13:05
    marekyggdrasil commented #2995
  • Oct 19 11:13
    marekyggdrasil commented #2995
  • Oct 19 11:11
    marekyggdrasil commented #2995
  • Oct 19 11:09
    marekyggdrasil commented #2995
  • Oct 18 07:29
    JosephGoulden commented #3084
  • Oct 17 13:17

    quentinlesceller on master

    mmr simplified Chinese translat… (compare)

  • Oct 17 13:17
    quentinlesceller closed #3095
  • Oct 17 13:17
    quentinlesceller commented #3095
  • Oct 17 10:34
    AurevoirXavier synchronize #3095
  • Oct 17 02:42
    AurevoirXavier commented #3095
  • Oct 17 02:39
    AurevoirXavier synchronize #3095
  • Oct 16 21:14
    JosephGoulden synchronize #3064
Ignotus Peverell
@ignopeverell
that seems like a lot of stale/rejected with a single miner
Ignotus Peverell
@ignopeverell
@yeastplume I think I have a fix, the kernel check needs only a kernel rewind, current code has booleans to allow that but still does a bunch of output-related stuff anyway
we just need to not do that extra crap (which is what fails), when all we're asking is a kernel rewind
Yeastplume
@yeastplume
ahh, yes makes sense... the input map isn’t really needed to rewind kernels then
Ignotus Peverell
@ignopeverell
indeed, requires a bit of deep surgery though because those things are still passed all the way down to the PMMR
Yeastplume
@yeastplume
I added those booleans during the messy merge to clean up some redundant single pmmr rewinds
Ignotus Peverell
@ignopeverell
yeah, good call
we just need to push that down a little further
or possibly have a specific kernel-rewind-only path, I'll see which seems cleaner
in the meantime we can just comment out the kernel history check
just a one-liner
Yeastplume
@yeastplume
heh, that’s what I did locally
that kernel check looks really, really inefficient .. hope we can come up with something more elegant at some stage
(just peanut gallery observation about code I haven’t been involved in)
Quentin Le Sceller
@quentinlesceller
@ignopeverell I had! China is a really interesting country :-).
Blade Doyle
@bladedoyle
@ignopeverell I see similar results....but Im not convinced the issue(s) are in the stratum server. I'll do some investigation to determine:
1) Whats wrong with the rejected shares - does the miner return invalid pow? - does the data get corrupted in transit? - is there some issue with the validation code?
2) What causes shares to be late - Is the grin node slow to build a new block template? - Is the stratum server slow to send it? - Is the miner slow to start on new jobs? - Something else?
I can volunteer to create github issues for these and start the investigation but will probably need to hand them off if the issue(s) end up being deep in the grin core code.
Quentin Le Sceller
@quentinlesceller
@ignopeverell regarding the versionned MMRs. I have been progressing really slowly the past weeks mostly about understanding the structure and its implications. However, I was planning to work all this week on Grin so I will try to produce something.
Ignotus Peverell
@ignopeverell
@yeastplume practically that check should be a couple order of magnitude faster than range proof checking but we'll see how it scales over time on T3
@bladedoyle that'd be great, thanks!
@quentinlesceller ok great, that'd be good to have
going to start the dev meeting in a few min
Quentin Le Sceller
@quentinlesceller
:+1:
Yeastplume
@yeastplume
@bladedoyle that's fine, I can look into any issues arising there as well
Blade Doyle
@bladedoyle
@yeastplume Yes please! ;)
Ignotus Peverell
@ignopeverell
alright! starting now
how's everyone doing?
Yeastplume
@yeastplume
awesome!
Quentin Le Sceller
@quentinlesceller
Great and you @ignopeverell ?
Ignotus Peverell
@ignopeverell
good, thanks
@yeastplume happy with the distraction from wallet code? :-)
Yeastplume
@yeastplume
there's only so much javascript one can take in the course of a career, even now
Ignotus Peverell
@ignopeverell
so we're basically getting ready for T3, all the consensus breaking stuff is in so all that's left to do is making sure it's all not entirely broken
given that we've prolly rewritten a third of the code since T2, or something close to that
so far things are going well, @yeastplume fixed a few mining things, I've been setting things up and testing sync and working on some more fixes there too
so I guess we're getting there
Yeastplume
@yeastplume
optimistically feeling that t3 will be more of a t3/beta than anything else
Ignotus Peverell
@ignopeverell
@yeastplume anything else you've spotted as being broken?
I'm thinking I'll send an email to @yeastplume and @quentinlesceller maybe in a couple days with a pre-T3 seed node IP so we can do a mini-test among a few of us
does that sound good?
Quentin Le Sceller
@quentinlesceller
Sure!
Yeastplume
@yeastplume
not broken outright with respect to consensus breaking... some tightening up yes.. only thing outstanding for me is difficulty adjustment and ensuring we can practice hard-forks easily on the next consensus breaking issue
yep, sounds like a plan
Ignotus Peverell
@ignopeverell
so we have one more day or two to iron out known kinks and then we can see if things still hold together with multiple machines
and send ourselves some pre-T3 grins too
@yeastplume regarding practicing hard-forks, hopefully @quentinlesceller will have progressed a bit on MMR versioning so we can have a better idea
and hopefully @tromp will pop out to check the diff adj or I'll take a look after I'm done with the MMR validation issue
Quentin Le Sceller
@quentinlesceller
Yes as I was saying earlier today I’m will be working on that this week
Ignotus Peverell
@ignopeverell
I'm hoping we can get real close by the end of the week
Yeastplume
@yeastplume
I'm messing with the diff adjustment as well locally, pretty sure we can get there
Ignotus Peverell
@ignopeverell
ok sweet