test-controller activity-hash check is too sensitive #3718
Comments
warner
added a commit
that referenced
this issue
Aug 18, 2021
The kernel "activity hash" is sensitive to everything written into the kvstore, including the bundled sources of built-in vats (like comms and vatAdmin), which include upstream libraries like marshal and assert. This makes the hash sensitive to a significant amount of code, and test-controller was asserting a specific value. This required updating the test's expectations on every code changes. Instead, we now merely check that `controller.getActivityhash()` returns a correctly-shaped string, instead of examining the actual value. closes #3718
warner
added a commit
that referenced
this issue
Aug 18, 2021
The kernel "activity hash" is sensitive to everything written into the kvstore, including the bundled sources of built-in vats (like comms and vatAdmin), which include upstream libraries like marshal and assert. This makes the hash sensitive to a significant amount of code, and test-controller was asserting a specific value. This required updating the test's expectations on every code changes. Instead, we now merely check that `controller.getActivityhash()` returns a correctly-shaped string, instead of examining the actual value. closes #3718
warner
added a commit
that referenced
this issue
Aug 18, 2021
The kernel "activity hash" is sensitive to everything written into the kvstore, including the bundled sources of built-in vats (like comms and vatAdmin), which include upstream libraries like marshal and assert. This makes the hash sensitive to a significant amount of code, and test-controller was asserting a specific value. This required updating the test's expectations on every code changes. Instead, we now merely check that `controller.getActivityhash()` returns a correctly-shaped string, instead of examining the actual value. closes #3718
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Describe the bug
@kriskowal made a minor change to some code (probably
assert) which gets folded into one of the built-in vats , and found thattest-controller.jsstarted to fail.As part of #3442 , we fold kernel state into this "activity hash", and that state includes the vat bundles for all the built-in vats, such as comms and vatAdmin. These vats, in turn, are built by calling
bundleSourceon code from the swingset package, but the bundling process pulls in a number of external libraries (includingmarshalandassert). So any changes in those upstream libraries will cause the initial activity hash to be different.Basically, the test is too sensitive. The plan is to downgrade it to merely assert that
controller.getActivityhash()works at all, and returns a hash-shaped string, but not assert the actual value.The text was updated successfully, but these errors were encountered: