Finally it worked. I've carefully read the automatingpost-review section,
but I misunderstood some permission.
I still don't think I understand it now, but at least, when I assign
everything to the logged user (who submits on be half of another user), it

Some suggestions:
1. The permission box is rather small, so actually I can't see the full
description of each permission (I can't scale it, nor can I scroll it
2. If post-review reports something like "xxx permission expected", it'll be
more helpful :)
3. I've noticed that the svn repository URL must match EXACTLY. for example,
if I write http://server/path somewhere but http://server/path/  somewhere
else, post-review will report something like "URL not matching".

BTW: still needs reporting backtrace? (just to confirm: you mean the
backtrace i've already attached?)

Now the whole process goes smooth manually. I'll start writing the automated
script tomorrow. Thanks again, Christian!

There is another question: is it possible to TRANSPARENTLY automate
pre-commit reviews? I mean, coders still use "svn ci" to commit, then
pre-commit hooks fetch the diff file, generate review request, then after a
reviewer accepts this diff file, it'll be automatically submitted to the
repository, on be half of the review requester. I found this theoretically
possible (e.g. store the diff in database first), but may lead to some
complexities (e.g. changes may finally arrive the repository in a different
order of their requests).

What's worse, pending submission may even be conflicting. For example
1. user A submits file X, waiting for review
2. user B submits file X, waiting for review.
Even though A's X and B's X are conflicting, step 2 will not get any error
since A's X did NOT arrive the repository. svn can only detect conflicts
between the working copy and a file in the repository history, but not

My question: what is the best practice (so far) that enables pre-commit
reviews? install RBtools on every client and force people to use post-review
to generate a request before submitting? I can't find a way to really force
them to do so :(

- Rujia

On Thu, Aug 20, 2009 at 6:18 PM, Christian Hammond <chip...@chipx86.com>wrote:

> Looks like we're not properly checking whether there's an error and
> assuming it's a draft. Certainly a bug. If you can file this backtrace in a
> bug report, that would be helpful.
> The problem is that the user specified doesn't have permission to modify
> that review request. The code that checks that is returning a
> WebAPIResponseError, but the code calling the function that received that as
> a return value isn't checking to see if it got an error response.
> Is the username and password you're using that of the user who submitted
> the code, or is it a special user with the special permissions? See
> http://review-board.org/docs/manual/dev/users/tools/post-review/#automatingpost-review
> Christian
> --
> Christian Hammond - chip...@chipx86.com
> Review Board - http://www.review-board.org
> VMware, Inc. - http://www.vmware.com

You received this message because you are subscribed to the Google Groups 
"reviewboard" group.
To post to this group, send email to reviewboard@googlegroups.com
To unsubscribe from this group, send email to 
For more options, visit this group at 

Reply via email to