You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The release process has a giant bottle neck in the form of going though each of the examples and updating it to the next version of Azle. I had an idea where we parallelize it by spinning each one up into it's own job, the result is committed to a unique branch, and then after all of those steps are finalized spinning up one more job that will go through each of those branches and merge them back into the release branch.
The initial prototype indicates that this in fact possible! So we want to make a new release process that will:
update candid
run benchmarks
update package.json dependencies
set Azle to the next version
make sure other dependencies are up-to-date like the agent, ts, etc
Actually lets just remove most of the dev dependencies so we don't have to worry about it
update package-lock with the new stuff from above
As long as we are at it we may want to look into updating Azle's:
package.json
Cargo.toml
corresponding lock files
Establish the baseline:
Make sure that all examples have run at least once before introducing the updates to rquickjs
bulk run all of the easy ones
hand run all of the fragile ones
Other things to do
Make a PAT and add it to the secrets with proper security
Add a workflow that will delete branches from failed release runs
Make sure that the benchmark.json files pretty print so we have some sort of nice formatting even before prettier
make sure that the benchmark output the full file path like testing
figure out why the broken ones are broken
PEM issues
make sure release works
make sure release does tests afterwards
Why did it merge the commit into the branch I was working on instead of the new branch that I made?
it did make the branch I think
Is it getting the version from the place? I am concerned that the version is always coming from the package.json even when it is recording the previous version. It just seems weird that most of the versions are same for baseline and current
main merges need to do npm only if its a release merge and repo if it's just a regular merge
we need to bring back the weird if statment that will keep the test from failing if there is nothing to commit
make sure the tests run after a automatic pr
The text was updated successfully, but these errors were encountered:
The release process has a giant bottle neck in the form of going though each of the examples and updating it to the next version of Azle. I had an idea where we parallelize it by spinning each one up into it's own job, the result is committed to a unique branch, and then after all of those steps are finalized spinning up one more job that will go through each of those branches and merge them back into the release branch.
The initial prototype indicates that this in fact possible! So we want to make a new release process that will:
As long as we are at it we may want to look into updating Azle's:
Establish the baseline:
Other things to do
The text was updated successfully, but these errors were encountered: