[snowblossomcoin/snowblossom] Issue closed by cluelessperson
SIP2 - My cat is pretty and smells nice
`vote_yes=1,2`
@Clueless I think I'll have MrPlow output a status report every few minutes.
It'll have miner rates 1min, 5min, 1hr
Pool rates Etc
That'd be great. For now I'm just regexing the log and reporting values.
Yeah, that is suck
If you could also report a total number of connected miners that'd be cool too.
because I was just about to count up the unique addresses from the log
sure
@Fireduck Let me number this properly. These are what I think we care about. 1. Total Rate. 2. Rate per miners/address 3. Number of addresses/miners 4. Announce when block is won
roger
If those are logged, I think it'd be perfect. :D
I have the webpage on a simple cronjob that updates the index.html with latest hash rate. I can have output the winners
rate reporting added to duckutil
now I need to get this baby to sleep
[snowblossomcoin/snowblossom] Issue opened by tster123
?
ah, I'm wrong
it's just a proof check fail
I thought it was the thing blocking
memfield_precache_gb=X this will move exactly X GB of the memfield into memory while using the rest from disk
[snowblossomcoin/snowblossom] #75 hybrid memfield
@Fireduck if mrplow could flush a status.txt (or xml or json) periodically, that’d be neat
a thought: voting blank?
distinct from not voting
@Rotonen What is voting blank?
@Clueless run that code with report_path set to something
and it will make you a sweet hash rate report
$ cat pool_report.txt Total: 1-min: 69.9K/s, 5-min: 14.0K/s, hour: 1.2K/s snowtest:emcpfwr7m5u6zmg3n4wmnz3lf4lz9h5ecarth54a 1-min: 69.9K/s, 5-min: 14.0K/s, hour: 1.2K/s
@Fireduck very nice - for when are you planning next release?
don't have a specific plan
most of the code changes have been to peer service which I imagine people are running from master branch
is there a method to tally votes yet by explorer yet?
will setup his own explorer
I started looking at that last night
ended up spending 2 hours figuring out why my local shackleton was throwing away every other request
If the binary name matches a directory name (case-insenstive) there is a build error: * * * protolib/protolib.pb_java.jar protolib/protolib.pb.descriptor_set * * * INFO: Analysed 9 targets (0 packages loaded). INFO: Found 9 targets... ERROR: /Users/jgleason/t/snowblossom.git/BUILD:38:1: Creating runfiles tree bazel-out/darwin-fastbuild/bin/Shackleton.runfiles failed (Exit 1) _bin/build-runfiles (args bazel-out/darwin-fastbuild/bin/Shackleton.runfiles_manifest bazel-out/darwin-fastbuild/bin/Shackleton.runfiles): mkdir '*main*/shackleton': File exists [17] INFO: Elapsed time: 0.568s, Critical Path: 0.36s INFO: 1 process, worker. FAILED: Build did NOT complete successfully A work around of changing: java_binary( name = "Shackleton", to java_binary( name = "ShackletonS", Made the build succeed. So something about a directory shackleton already existing. Not sure if that is because of the class of that name or the source directory or what. Not sure what to do about it either. Maybe if we make the binary ShackletonExplorer.
i’ll need to bother figuring out flushing stats to a nicer looking site with a static site generator eventually, but gonna roll with actual releases - i might be old fashioned
I think we should make a handful of minor improvements and then roll out a simple quality of life update
sounds fair
less repetitive logging would be one of those for me
@Fireduck @Tyler Boone Do you think it would be good to setup an automatic release schedule?
Create a policy that it doesn't go into the release unless it's dire or has a week of testing?
I can setup test machines to build nightly to test changes. We monitor them for errors in logs, block height, etc.
and I can setup a test-production, one that immediately builds new releases and runs them
same purpose
@Clueless as far as i can see it's 2 devs, a few enthusiasts - no need for red tape
agreed.
building systems is enjoyable, but not the purpose itself
wrong hammer, and with enjoyable hammers everything looks like a nail, etc.
@Rotonen Another thought is that I can setup a node that automatically rebuilds **every commit* so we can more readily test changes.
rather set up integration testing via travis
i suppose step 1 would be 'will it still build', namely i still cannot build shackleton
also the build is a bit noisy on a modern clang on macos, but nothing critical in the way of warnings, but i could look at rigging scan-build for the c++ bits
no idea what you're saying.
I'd just automate away building constantly, and if it errors, alerts us so we can check the log and fix it
There is an issue for Shackleton. Case sensitive issue
yeah, just use travis and/or appveyor for stuff like that, straight up github integrations and free for open source
Opened today
https://clang-analyzer.llvm.org/scan-build.html <- also scan-build is just about my all-time favourite toy, so i'll try jamming that about anywhere i touch
IMO, automatic builds and releases are time savers, but setting them up is work
a simple build + smoke test via travis should not be that hard to pull off
elegance can follow over time
@Fireduck The reporting is niceeee.
We have some solid integration tests already
@Rotonen I'll look into travis. :I I'm not averse, just ignorant.
Just need more of then
Them
@Fireduck how do you run the tests?
bazel test //...
thank you
and that seems to build :all as well, so i cannot run those, as shackleton craps up, but that'll eventually settle
but neat
shackleton is broked?
@Fireduck hinted it'd be some case sensitivity issue and that'd make sense as i'm on APFS
Read the issue
but yeah, test obviously requires build :wink:
and some devs are on NTFS
whoops
earlier there was also a classic git defaults vs. line endings issue
trying to figure out bazel: is there an easy way to build all the deploy targets at once, like some built-in meta tag, or would that have to be defined in the BUILD file?
Not sure
@Rotonen Can you confirm discord user: Rotonen#7301 is you?
@Clueless yeah, why?
@Rotonen figured I'd make you mod.
so long as you do not expect me to actually put forth any moderation effort
No expectation. Just seem reasonable and not powermad
@Fireduck Perhaps a nonsense question. Would it ever be possible to point snowblossom processing at research data? curious (and maybe dumb)
boinc already exists and they're good at that
My thinking is that then the data fields are self-generating, have no limit, it would shut idiots up about power usage, and it would buy us "helping mankind" votes.
Snowblossom would not be useful for anything
I figured it was probably a stupid premise