rebase

rebase

Various nits

braces

At least for the field size thing.

This doesn't really answer the question.

No. This makes history ungreppable.

Ok so you you ave something that goes roughly as

I' not sure why you would revert back other changes, there were not bad per se.

In D7475#177718, @PiRK wrote:

update diff message with arc diff --edit

Please fix the test plan, it doesn't display properly.

This is doing exactly the wrong things. Instead of consolidating the framework in which automated commit happen, this instead consolidate the fact that there some kind of big ball of mud type script that can do every signle automated commit, but only half the workflow.

You want something like:

This does not solve the problem mentioned, because there might be more changes in between the rebase and the push. This is fundamentally broken and cannot be fixed with this approach.

I don't see what problem this solves. The description doesn't explains this.

run_silent is a pretty bad name, considering this is not running "silent". Running something silently is also not the only thing that this script is doing.

How come arc lint didn't yell at us without this patch? Not really a change request, but this needs to be figure out.

There are manpages in tree, you probably want to clean this up.

You probably want to make the makepage generated via the build instead of being committed in source control.

Accepting, but this is looking more and more like you want to invest in some python here.

You are build a Rube Goldberg machine. Just have some kind of automated patch script in which you can plug whatever code generation part you want.

The title is wrong.

The description doesn't really explain the problem that this solves or why this approach. For instance why do we need to send the patch to phabricator to then redownload it at the other end? That seems to provide zero value over transferring the diff right away.

lol