Hi,

I have problems with using RB correctly.

What I did do and need to post for review.

1. I cloned master, created a local "feature" branch, worked, worked, etc...

*eventually* ending up with 3 commits on my local branch:

master < commit1 < commit2 < commit3

2. I called `rbt post commit1` (of course, using sha1)

rbt reported success and printed URI, which I checked and confirmed using browser.

3. I called `rbt post commit2` *before* commit1 was reviewed

rbt failed with "The file was not found in the repository (HTTP 400, API Error 207)" and left an "empty" review (i.e. without actual diff) in RB.

Does this mean that RB depends on having the parent already pushed? Because if that's true, I can only think of bad solutions:

* Either [A] wait (+hassle reviewers) with posting commit2 until after commit1 is reviewed + pushed to origin/master,

* or [B] squash it all and post it as one huge review,

* or [C] don't squash it but still post it as one huge review (specify range of commits to RBT),

* or [D] base all commits on origin/master,

* or [E] have a "relay" branch in a repo accessible by RB (either main repo or "personal" per-user repo for WIPs).

(Please tell me I'm wrong and correct me.)

What I've been advised and tried:

* "try again after a while" ... I've tried ~5 times, then once next day ... no change

* "open the empty in RB and upload the diff manually" ... but I ended up with the similar error: the commit that my diff decalres as parent does not exist

* "look at script by vcrhonek" ... but I didn't try it since I'm not sure what it actually does

From looking at code it seems to do the [C] but I don't want that since in my case I think it would make the review PITA since e.g. one commit is moving code, other is changing it and yet the commits aren't functionally relevant.

So: What is the right way? Or a known workaround? Or is my workflow silly?

Thanks, aL.

P.S.: If you want to see the commits, it's the last 6 commits here:

http://git.engineering.redhat.com/git/users/amahdal/openlmi-providers.git/lo...

The first one has already been posted (/r/1684/)