Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 18:36
    arnetheduck commented #483
  • 18:34
    arnetheduck commented #481
  • 18:34
    arnetheduck closed #481
  • 13:59

    arnetheduck on master

    simulation: make default setup … (compare)

  • 12:03

    arnetheduck on master

    readme: simplify prerequisites … (compare)

  • 11:57

    delete-merged-branch[bot] on ter

    (compare)

  • 11:57

    arnetheduck on master

    implement beacon_reorgs_total m… (compare)

  • 11:57
    arnetheduck closed #491
  • 04:40
    oskarth commented #406
  • Oct 21 22:01
    stefantalpalaru synchronize #488
  • Oct 21 22:01

    stefantalpalaru on CI

    Travis: go back to Ubuntu Bionic (compare)

  • Oct 21 21:43
    stefantalpalaru synchronize #488
  • Oct 21 21:43

    stefantalpalaru on CI

    Travis: ARM64 build Azure chan… (compare)

  • Oct 21 18:00
    tersec opened #491
  • Oct 21 17:59

    tersec on ter

    implement beacon_reorgs_total m… (compare)

  • Oct 21 15:46
    kdeme edited #177
  • Oct 21 15:45
    kdeme edited #177
  • Oct 21 15:45
    kdeme edited #177
  • Oct 21 15:45
    kdeme edited #177
  • Oct 21 15:45
    kdeme edited #177
Autobot
@status-im-auto
zahary@discord: We haven't prepared much in terms of end-user documentation yet, but you can find build instructions for our node here:
https://github.com/status-im/nim-beacon-chain/
IgnorantFoolio
@IgnorantFoolio
Is there a public testnet up and running for the latest ARM64 builds?
Ștefan Talpalaru
@stefantalpalaru
No separate testnet, we're using the common one. Wait a bit for ARM64 support for nim-beacon-chain. I'm not done yet fixing a BearSSL wrapper problem.
IgnorantFoolio
@IgnorantFoolio
@stefantalpalaru Thanks for the update.
Autobot
@status-im-auto
cheatfate@discord: @stefantalpalaru i can help you in fixing bearssl
Ștefan Talpalaru
@stefantalpalaru
I already did it: status-im/nim-bearssl#2
Ștefan Talpalaru
@stefantalpalaru
The macOS backlog is spiking like crazy on Travis: https://www.traviscistatus.com/#week
Autobot
@status-im-auto
cheatfate@discord: @stefantalpalaru ARM64 LE or BE?
Ștefan Talpalaru
@stefantalpalaru
Probably LE, since this is the default.
Ștefan Talpalaru
@stefantalpalaru
Why is Chronicles complaining in the nim-beacon-chain test suite?
[Chronicles] A writer was not configured for a dynamic log output device. Log message not delivered: NOT 2019-10-17 19:13:28+02:00 Exit: not in validator set long enough tid=28125
Ștefan Talpalaru
@stefantalpalaru

Weird Nim error in Azure, after bumping nim-libp2p and maybe making some path longer in the process - https://dev.azure.com/nimbus-dev/nim-beacon-chain/_build/results?buildId=194 :

[NimScript] exec: nim c --out:./build/test_fixture_ssz_static -r -d:release -d:chronicles_log_level=DEBUG -d:const_preset=minimal --verbosity:0 --hints:off --warnings:off tests/official/test_fixture_ssz_static.nim 
Error: invocation of external linker program failed. The filename or extension is too long.

Similar bug report: nim-lang/Nim#6452

Ștefan Talpalaru
@stefantalpalaru
Could it be that "nim-bearssl/bearssl/decls.nim" passes absolute paths to the compiler/linker and in combination with Azure's current working directory it somehow goes past the already raised path limit?
Autobot
@status-im-auto

zahary@discord: The Chronicles error is coming from a modification in nim.cfg that's needed only in the beacon node. I shouldn't have put it in the global top-level nim.cfg file.

The log output in the beacon node is now written in a slightly special way to accomodate the added status bar at the bottom

Ștefan Talpalaru
@stefantalpalaru
Oh, OK.
Autobot
@status-im-auto

zahary@discord: Thanks to @stefantalpalaru's latest fixes, we are getting closer to being able to switch to Nim 1.0.2. I've now also updated the beacon-chain PR here:
status-im/nim-beacon-chain#489

... but we have a unit test failure on Windows 32 bit. Who was able to reproduce problems in this environment? @jangko?

Ștefan Talpalaru
@stefantalpalaru
"The maximum command line length for the CreateProcess function is 32767 characters." - https://devblogs.microsoft.com/oldnewthing/?p=41553
The one used for linking is 33987 :-)
andri lim
@jangko
@zah I can reproduce the failed interop test on windows 32 bit. now identifying whats wrong.
andri lim
@jangko
andri lim
@jangko
last blocking issue with Nim v1.0.2 just fixed. hopefully the migration process will be smoother.
Ștefan Talpalaru
@stefantalpalaru
Nimbus devel branch merged into master and deleted. PR targets changed accordingly.
Who's doing the Nim migration?
BlockedHouse
@BlockedHouse_twitter
hey guys i heard on townhall meeting while back about zilliqa being discussed, do we have update on that
Autobot
@status-im-auto
zahary@discord: How do you feel about switching to Nim 1.0.2 today? Do we have any reasons to delay it a bit more?
andri lim
@jangko
I think its fine to do it today
Ștefan Talpalaru
@stefantalpalaru
Sounds good.
Ștefan Talpalaru
@stefantalpalaru
New Travis ARM cache up and running, but unfortunately buggy: I'm getting on ARM the cache I set on AMD64.
Waiting to see if they want to fix it on their side before making ARM-specific cache dirs on our side.
Autobot
@status-im-auto
mratsim@discord: @zahary @jangko I see that the 1.0.2 PR is ready, merge?
Autobot
@status-im-auto

zahary@discord: Yes, I was waiting for more people to chime in yesterday. I think I have to push one final fix to Confutils in that branch and I'll merge it

After merging, it should be enough to do

rm vendor/nimbus-build-system/vendor/Nim/bin/nim
make update deps

And the first line is "just in case". You can run it only if the second one doesn't start rebuilding Nim

Dustin Brody
@tersec

Current profile (with -d:releaseon the testnet):

    26.66%  beacon_node  beacon_node           [.] sha256Transform_H0389bySRtxG8VYJBMMV8Ag
    19.92%  beacon_node  beacon_node           [.] BIG_384_58_monty
     9.42%  beacon_node  beacon_node           [.] BIG_384_58_mul
     3.43%  beacon_node  beacon_node           [.] BIG_384_58_mod
     3.42%  beacon_node  beacon_node           [.] BIG_384_58_sqr
     1.99%  beacon_node  beacon_node           [.] finalize256_eVWav0d2UD0xBevbiUEQUA_2
     1.71%  beacon_node  beacon_node           [.] Marker_tySequence_iPL3rW9a9bV9cIzaXekL6AsDg
     1.61%  beacon_node  beacon_node           [.] BIG_384_58_fshr
     1.36%  beacon_node  beacon_node           [.] interiorAllocatedPtr_N7XNyRgfMLnCjS9bVeVc3bA
     1.31%  beacon_node  beacon_node           [.] BIG_384_58_jacobi
     1.25%  beacon_node  beacon_node           [.] genericAssignAux_mQY3YYo2RBkhPz09ajfLwpQ
     1.22%  beacon_node  beacon_node           [.] rawNewObj_BpNS5yt3b9cMOponsUKh8aw
     1.18%  beacon_node  beacon_node           [.] BIG_384_58_norm
     1.04%  beacon_node  beacon_node           [.] rawAlloc_yn9c8RLaS8vgVBeMBfmkdUg
     1.03%  beacon_node  beacon_node           [.] sweep_zoTIuavlrCHyw3B8MBaLJA_5
     1.02%  beacon_node  beacon_node           [.] genericReset
     0.88%  beacon_node  beacon_node           [.] BIG_384_58_add
     0.81%  beacon_node  beacon_node           [.] BIG_384_58_dnorm

So, dominated by hashing, and probably quite repetitive hashing, and crypto, some of which might be avoidable but isn't used as wastefully as hashing is in nim-beacon-chain.

There's a disconcerting amount of memory allocation, just directly showing up, across genericAssignAux, rawNewObj, rawAlloc, genericReset, etc
And if 5% or so is directly attributable to such functions, the GC is being pushed quite hard
Autobot
@status-im-auto
mratsim@discord: Any idea if the memory alloc is related to a particular subsystem (async or ssz for example) or maybe compute (sequence of validators being allocated/destroyed)?
Dustin Brody
@tersec
@mratsim at least the genericAssignAux seems to mostly show up in state_transition
i.e. copying BeaconState is expensive. but that was already known, e.g., @zah's status-im/nim-beacon-chain#484
genericResetAux generally shows up in the context of initDeque, in turn in the context of initWithSinglePage
Which in turn is from newSszHashingStream, so, genericResetAux is SSZ specifically
Autobot
@status-im-auto
arnetheduck@discord: wow, these are pretty nice: https://proto.school/#/
arnetheduck@discord: re 1.0, looks like a point of no turning back - after it, we should probably take note of any debt/workarounds that we've built up over the months and clean that up too. moving forwards, we could simplify our life by supporting only 1.0.
arnetheduck@discord: I'm also somewhat uncomfortable with the potential disuption it might cause, compared to more pressing needs that are actually on our roadmap, but it seems that it'll keep distracting from those until we merge it
Andrew Quentson
@Aquentson_twitter
Hi, are we expecting the testnet to launch anytime soon? Sounds like next month at best now?
Autobot
@status-im-auto
proto@discord: @mratsim @zahary the ssz repo thing is very much a draft right now. Goal is to express details better than what's possible in the eth2 specs repo, and add the newer things. And implement some feature staging and change control; I don't want partials and merkle proof backings to live in a PR forever.
More work to follow tomorrow, and will share at the call. And sharing more about a new test service, going to hook up nimbus with ncli :)
Autobot
@status-im-auto
mratsim@discord: btw feel free to use the eth2.0-clients organization as you see fit, I was in vacation just after Devcon but I will move the client scripts from nim-beacon-chain tomorrow, we can probably have another repo with all the *cli build scripts so that it's more easy to consume and updates for your test/fuzzing needs
Autobot
@status-im-auto
arnetheduck@discord: status-im/nim-libp2p#10
dryajov@discord: ethresearch/p2p#4
Autobot
@status-im-auto
arnetheduck@discord: libp2p/specs#195