363237 - bootstrap Update step should not run independent verification

Status: RESOLVED INVALID

(Release Engineering :: General, defect)

Description

[Robert Helmer [:rhelmer]]

Full verification should be done in a separate step, since it needs to be run after snippets are deployed. This patch removes the full, independent update verification, and leaves the quick "only test channels" verification.

The full verification will be reinstated later, in a different patch.

Comment 1

[Robert Helmer [:rhelmer]]

Attachment: remove full/independent verification from Update step

Comment 2

[J. Paul Reed [:preed]]

Comment on attachment 248040 [details] [diff] [review]
remove full/independent verification from Update step

This is fine; you said you still had this code in your personal branch, so as long as we can easily resurrect it, I'm cool with deleting it wholesale.

Comment 3

[Robert Helmer [:rhelmer]]

Attachment: add standalone Verify step, plus push/announce support

This adds a standalone Verify step, plus Push/Announce support. I can split the latter out if you like, but it's pretty minor so I'd like to get it in as long as I'm touching "release".

Comment 4

[Robert Helmer [:rhelmer]]

Actually, going back to having this as a verification step for Update instead of a standalone step. It'll be short-term pain, but it'll make more sense in the long run when we can kick of builds and verifications remotely.

Comment 5

[J. Paul Reed [:preed]]

Comment on attachment 251298 [details] [diff] [review]
add standalone Verify step, plus push/announce support

r- because this patch still has the separate Verify step in it.

Also, for the Push() and Announce() sub-classes that are effectively null, I'd either remove the Log() messages entirely, or say something about "No announce behavior defined."

Comment 6

[Robert Helmer [:rhelmer]]

(In reply to comment #5)
> (From update of attachment 251298 [details] [diff] [review])
> r- because this patch still has the separate Verify step in it.

Yeah, I guess changing the resolution of the bug to INVALID does not remove review requests, sorry :(

> Also, for the Push() and Announce() sub-classes that are effectively null, I'd
> either remove the Log() messages entirely, or say something about "No announce
> behavior defined."
 

Ok, I think removing them completely is best.