Now that we properly initialize governance, a ton of governance slashing
doesn't happen in the first few blocks. Because of this, we can run through
blocks in the range (0,200) quite rapidly. This PR acknowledges that and
increases many of the default block heights.
The intent of this is to allow for simulating things like slashing for not
voting on a governance proposal. To test this, you would queue all the validator votes
in future blocks, and keep track of which ones you didn't slash. Then you could add queue a
"check governance slashing operation" after the voting period is over.
Note each defer occurs a 30ish ns overhead here, which is significant
for IAVL iteration. We were previously using around 3-4 defers. (2 in
next, one in Valid, one in Key, one in Value) This slows down the entire
application quite significantly, as we require fast iteration.
This commit makes it such that circle CI only shows the module whose
tests it is currently running in the log, unless a test fails. For each
failing test, it will display all of the log output. This is done to make
our log output far more scrollable. We lose no information in debugging.
Additionally I've added the following entries:
* Added MinInt and MinUint functions
* The keys sub-module is now in the sdk
* [keys] New keys now have 24 word recovery keys, for heightened security
* \#872 - recovery phrases no longer all end in `abandon`
* Adds a Min function to Int, and uses that in the slash function
* Adds a getHeight helper function to iavlstore
* Adds a splitPath function to baseapp
* Changes cyclo param from 10 to 11
Gocyclo is a code complexity linter. It uses cyclomatic complexity.
Cyclomatic complexity essentially measures the number of different
paths code could go through. (The conditional in a for loop counts
as adding one path) It looks at this on a per-function level. The
idea that this would be enforcing is that if there are too many
different paths code can go through in a function, it needs to be
better split up. (A function with too many code paths is hard to
reason about)
The complexity which we want the linter to start failing on is
configurable. The default is 10. Change the "Cyclo" parameter in
`tools/gometalinter.json` to try other values.
* `gaiad init gen-tx` makes the outputted file use bech32, with acct prefix
* `gaiad init --gen-txs` only reads bech32 with acct prefixes
The reason for using the account prefix is that in principle you could
have genesis transactions for non-validators.
Closes#1475
* Makes all cli tests use .test_gaiad, .test_gaiacli instead of the
same directories as the default binaries
* Abstracts alot of the functionality for setting up the server into
a single function / file-wide constants. This is to reduce
code duplication, especially since some of this functionality
depends on each test setting up the keys in the same way.
Closes#1461
Now provides better error messages when the account you're sending
from has no money, or it has insufficient funds. (Avoids making
the user interpret ABCI errors)
closes#1489
ValarDragon