fix: go mod would not copy the non-go files by default

[andyliuliming]

Reason for Change:

fix: go mod would not copy the non-go files by default, so do go-bindata here is better. which will help the projects which vendored aks-engine .

Issue Fixed:

Requirements:

• uses conventional commit messages
• includes documentation
• adds unit tests
• tested upgrade from previous version

Notes:

[acs-bot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: andyliuliming
To fully approve this pull request, please assign additional approvers.
We suggest the following additional approver: serbrech

If they are not already assigned, you can assign the PR to them by writing /assign @serbrech in a comment when ready.

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[mboersma]

@andyliuliming can you make sure your commit messages follow https://www.conventionalcommits.org/? Otherwise they don't show up in the changelog we generate.

[andyliuliming]

@andyliuliming can you make sure your commit messages follow https://www.conventionalcommits.org/? Otherwise they don't show up in the changelog we generate.

done, thanks.

[mboersma]

We have tried before to check the go-bindata-generated files into version control, but found it created a git rebase nightmare. We reverted that change in #546. I don't think we want to do this again.

The tradeoff is that you have to run make build or make generate after go get github.com/Azure/aks-engine, but that's the compromise we chose.

[jackfrancis]

What @mboersma said is correct. We can't accept this change as-is.

[andyliuliming]

but if we switch to use the go module.
the go mod vendor will not copy the parts files for us.
so the go generate will fail at the repos which try to vendor aks-engine :(
@jackfrancis @mboersma

[mboersma]

So #1077 won't work without this change?

Theoretically, the binary data could be broken into small segments to ensure there was minimal contention over changes. If there were a very close mapping between the script or code and its go-bindata representation, then real-world merge conflicts on PRs would be meaningful.

But I doubt we would want to make things that complicated—if that approach would even work. In practice, checking in these files required every PR in the queue to be rebased every time one was merged.

[andyliuliming]

So #1077 won't work without this change?

Theoretically, the binary data could be broken into small segments to ensure there was minimal contention over changes. If there were a very close mapping between the script or code and its go-bindata representation, then real-world merge conflicts on PRs would be meaningful.

But I doubt we would want to make things that complicated—if that approach would even work. In practice, checking in these files required every PR in the queue to be rebased every time one was merged.

then one no-compress option will help :)

re-send one PR: #1088