How do you review code in squeak?
Because we could have another stream and push between one and the other.

I mean practically?
But this is because that the package arrives in MCBrowser that you read it?
How do you deal with changes requiring a lot of changes in multiple packages?
How do you know that they all work together?

Now I often have to fix some more changes when I integrate fixes because it 
happens that people
forget a detail there and there. 

Stef

> Hi,
>> 
>> We started to improve the states defined in the tracker... this now has:
>> 
>> 
>> FixToInclude         = QA has verified that the fix worked, needs to be 
>> integrated
>> FixReviewNeeded      = Code is there, but a review is needed
>> NoSourcesAvailable   = fix proposed, but machine readable code missing
>> Workneeded           = Next action is defined but not yet completed
>> NextActionNeeded = What is the next action?
>> Accepted             = Problem reproduced / Need acknowledged
>> New                  = Nobody looked at this yet
>> OnHold               = We will come back later to this issue, no action 
>> possible now
>> 
>> 
>> The issue tracker by default sorts the items to show the ones on top that 
>> have progressed
>> the most:
>> 
>>        http://code.google.com/p/pharo/issues/list
>> 
>> 
>> The idea is that the "FixToInclude" status should be integrated very quickly 
>> (within a day),
>> later maybe even automatically.
>> 
>>        Marcus
>> 
> 
> I think this is a good pragmatic list.
> 
> One thing I wonder is whether we can mark our own fix as "FixToInclude"
> In theory, we should not because we are bypassing a peer review.
> We should not lower quality insurance.
> But practically it depends on two things:
> - the level of work supported by integrators
> - whether you have enough peer reviewers
> and how easy or involving is such a review (the reviewer throughput)
> 
> I suppose Marcus would like to be replaced by an automaton and reduce
> integrator support.
> It shall then be mandatory to tag our own fixes as "FixReviewNeeded".
> But if ever you don't have enough reviewer throughput, then the
> "FixReviewNeeded" will start accumulating in the tracker.
> 
> If we fail to increase the throughput (with an army of reviewers or
> some facilitating tools) then Marcus (and Stef) burden won't lighten.
> 
> I repeat once again, but I prefer the squeak-trunk smooth process were
> the changes are published in a mailing list.
> They CAN BE reviewed more easily, and I'm sure they ARE reviewed more
> efficiently than in Pharo process (I mean by many more eyes).
> As a committer, I can then publish directly in trunk without the inbox stage.
> It's very like a direct "FixToInclude"
> But that does not mean there will be no review.
> But maybe I'm tainted ;)
> 
> Nicolas
> 
>> --
>> Marcus Denker -- http://marcusdenker.de
>> 
>> 
>> 
> 


Reply via email to