move to new protocol structure #179
Conversation
Fixes
File "/cbc-casper/simulations/network_delay.py", line 31, in gaussian_delay
return round(random.gauss(MU, SIGMA))
TypeError: 'int' object is not callable
simulations/network_delay.py
Outdated
| def gaussian_delay(sender, reciever, round): | ||
| """Messages take a random amount of time to deliver | ||
| (from a gaussian distribution)""" | ||
| return round(random.gauss(MU, SIGMA)) |
There was a problem hiding this comment.
@KentShikama Thanks!
Make math.round in gaussian delay no longer shadowed
naterush
force-pushed
the
chore/protocol-rework
branch
from
1892178 to
8de517b
Compare
naterush
force-pushed
the
chore/protocol-rework
branch
from
43a0ed2 to
d2e6fab
Compare
naterush
force-pushed
the
chore/protocol-rework
branch
from
74559a2 to
d6ac26a
Compare
casper.py
Outdated
|
|
||
| For more information, | ||
| join the gitter: https://gitter.im/cbc-casper/Lobby | ||
| read the wiki: |
casper.py
Outdated
| save=args.save, | ||
| # not all parameters are needed to generate the json for execution | ||
| execution_string = generate_json(**{k: v for k, v in vars(args).items() if k not in NOT_NEEDED}) | ||
| print(type(execution_string)) |
There was a problem hiding this comment.
I changed it to print the execution string - which I think we want (so you can duplicate the execution if you want).
Not sure why I was printing the type :)
|
|
||
| self.plot_tool = plot_tool(display, save, self.global_view, self.global_validator_set) | ||
|
|
||
| self.unexecuted = execution_string |
There was a problem hiding this comment.
I like the way one string is consumed by the other :)
casper/protocol.py
Outdated
| if message_hash not in validator.view.justified_messages: | ||
| current_message_hashes.add(message_hash) | ||
|
|
||
| while any(current_message_hashes): |
There was a problem hiding this comment.
Can't we just do
messages_needed = set()
current_message_hashes = set(message.hash)
while any(current_message_hashes):
etc...Instead of doing the first for loop which essentially repeats some code from the while?
There was a problem hiding this comment.
I think so! The duplicated code bothered me as well...
|
|
||
| return messages_needed | ||
|
|
||
| def execute(self, additional_str=None): |
There was a problem hiding this comment.
I'd be interested in seeing a step method that allows you to consume one token at a time, and just have the execute use the step until the string is totally consumed. Would allow for very granular inspection into what is going on in a protocol execution if needed.
Might be premature until we have a particular use case.
| self.children, | ||
| self.latest_messages | ||
| ) | ||
| print(available_outputs) |
There was a problem hiding this comment.
or is this part of the plot? If it is, move to plottool
|
|
||
| if fault_tolerance > 0: | ||
| self.bet_fault_tolerance[latest_message] = num_node_ft | ||
| if self.view: |
There was a problem hiding this comment.
Good catch. This was an issue in a previous version of the architecture, but I fixed it. Removing!
| def generate_rrob_execution(num_validators, num_rounds, network_delay_func): | ||
| curr_creator = 0 | ||
| def rrob_creator(): | ||
| nonlocal curr_creator |
There was a problem hiding this comment.
Nice, never seen nonlocal before
simulations/exe_str_generator.py
Outdated
| @@ -0,0 +1,78 @@ | |||
| """The simulution utils module ... """ | |||
tests/simulations/test_network.py
Outdated
| CONSTANT | ||
| ) | ||
|
|
||
| def test_meets_interface(delay_function): |
There was a problem hiding this comment.
two new lines between function defs
|
@djrtwo Thanks for the review! Addressed your comments. I'm going to merge this and then do a release (going to skip the sharding release). |
Based on spec in this issue: #175. Not finished yet, but opening for initial feedback.
For now, removing code for experiments and analysis. Will reintroduce this in future PR to be compatible with new protocol classes.
TODO: