Current status of JIRA workflow?

75 views
Skip to first unread message

Andy Fingerhut

unread,
Feb 9, 2012, 2:16:04 PM2/9/12
to cloju...@googlegroups.com
So I'm aware of these three pages of instructions and descriptions of workflow for Clojure tickets:

[1] http://clojure.org/patches

[2] http://dev.clojure.org/display/design/JIRA+workflow  - seems to have a lot in common with the link above, at least as far as how to create properly formatted patches, and perhaps more.

[3] http://dev.clojure.org/display/design/Clojure+workflow+%28experimental%29  - Given its name, I am guessing this is something that is proposed for wider use by Clojure and its new contrib modules, but is so far only implemented in one JIRA project as a demonstration?


Is [2] the most "current", and if it has any errors in it, the documentation ought to be changed to match the actual steps that people need to take on the JIRA web pages?

Thanks,
Andy

Andy Fingerhut

unread,
Feb 19, 2012, 2:19:48 AM2/19/12
to cloju...@googlegroups.com
In rereading my message, I probably didn't make it clear that my intent is that I would update the instructions to be current, after figuring out which of these I should start from, and which the Clojure developers would like to be the "home" for this info.

Is [2] a good home for up to date instructions?

Is someone already working on updates, and I should coordinate with them?

Has any decision been made on whether the workflow should change to what is described at [3]?

Thanks,
Andy

Stuart Sierra

unread,
Feb 22, 2012, 5:40:19 PM2/22/12
to cloju...@googlegroups.com
Is [2] a good home for up to date instructions? (http://dev.clojure.org/display/design/JIRA+workflow)

Yes. That's what I refer to when screening tickets. (Every. Damn. Time.) 
 
Is someone already working on updates, and I should coordinate with them?

I am not aware of anyone else working on this since Chas Emerick designed the experimental workflow.
 
Has any decision been made on whether the workflow should change to what is described at [3]?

Not yet. I think people are open to the idea, just haven't had time to look at it. To apply it to Clojure, Rich and/or Stuart Halloway would need to give the go-ahead. They'll be preoccupied at least until Clojure/West in March, but I'd like to bring workflows up again after that.

-S

Andy Fingerhut

unread,
Feb 23, 2012, 6:26:30 AM2/23/12
to cloju...@googlegroups.com
I've been looking at [2] and thinking about it, and coming to conclusions that might be similar to those that led Chas to create [3].

In the chart of [2], it would be nice if a ticket could advance at least up to the point of "Test" (i.e. ready to be screened, and a patch accepted or rejected), without having to be touched by someone who is a screener or in Clojure/core.

The current flow seems to require a Clojure/core member to touch a ticket twice to get it to "Test": once to Vet the ticket, and once to mark it as "Test" after a contributor has added a patch (or three times if it was Vetted in Backlog).  Why?  Because the Approval setting of a ticket is only supposed to be changed by a Clojure/core member (even though it appears that JIRA doesn't stop others from changing it).  That sounds like busy-work, but I may be missing of the reasons for the current flow's structure.

I don't know if the auto-generated chart of Chas's proposed flow at [3] is up to date, but if so, I've added another easier-to-read figure that should be equivalent to it on that same page.  His auto-generated chart does not explicitly state who can make which state changes, but I've added my guess/proposal to that chart.

Notice how it is like a funnel.  Anyone can create a ticket.  Anyone contributor can get it to the point of "Ready to Screen" (corresponding roughly to "Test" in the diagram at [2]).  Only then does it require one of the few, the brave, the Marines --- uh, I mean the screeners, to advance it further.  And if the issue is real and the patch is good, they only need to touch it once to get it to Rich.
--
You received this message because you are subscribed to the Google Groups "Clojure Dev" group.
To view this discussion on the web visit https://groups.google.com/d/msg/clojure-dev/-/PJwSWVycm7sJ.
To post to this group, send email to cloju...@googlegroups.com.
To unsubscribe from this group, send email to clojure-dev...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/clojure-dev?hl=en.

Stuart Sierra

unread,
Feb 23, 2012, 8:14:30 AM2/23/12
to cloju...@googlegroups.com
I like where this is going, Andy. "Vetting" has been slow, often ignored. The original idea was to have at least one person -- other than the original reporter -- verify that the reported issue is real. Ticket voting, which we've also discussed, might be another way to achieve this.

-S

Chas Emerick

unread,
Feb 23, 2012, 8:39:09 AM2/23/12
to cloju...@googlegroups.com
Despite the sad appearance of the diagram of the experimental workflow, it should still be an accurate reflection of what is described in the JIRA workflow.

I don't know if it's been mentioned in this thread, but there was a prior discussion about this workflow here:


When I left the conversation, there was a tentative "ok" from Stuart H. that something like this would be adopted given the resolution of two TODO items, pointed out in my message from Nov. 21.  Andy, feel free to drive it home if you like; I'm far too preoccupied with work around Leiningen and Counterclockwise of late.

Let me know if there are any other open questions that aren't clarified in the above thread or in the wiki page.

Cheers,

- Chas

Andy Fingerhut

unread,
Feb 23, 2012, 12:57:32 PM2/23/12
to cloju...@googlegroups.com
Chas, thanks for the pointer.  I had not read that discussion before.

While reading it, it appears that you may have created the state transition screen shot and published it at [1], and then later made some changes to the workflow (e.g. renamed "Rejected" as "Declined" and added transitions from everywhere to Approved and Patch Screened states).  Would you or someone else with JIRA admin access be willing to privately send me a screenshot of the current version?  I can then update the diagrams on the Wiki at [1].


Andy

Andy Fingerhut

unread,
Feb 23, 2012, 10:11:08 PM2/23/12
to cloju...@googlegroups.com
In thinking and playing around with JIRA filters today, I have a suggestion that might be useful.

[1] is a single-page summary that shows a proposed workflow that I believe we could adopt without any changes to JIRA at all.  The JIRA filters mentioned there have been created.  See [2] for instructions on how to find them.

Take a look at it and see what you think.

Andy



In the box labeled Author, type in Andy Fingerhut.  Click on the choice that pops up and the contents should change to my user account name, jafingerhut.  Click on the Search button at the bottom.

Chas Emerick

unread,
Feb 24, 2012, 7:09:25 AM2/24/12
to cloju...@googlegroups.com
Andy,

You're right.  I just replaced the prior diagram with one generated by JIRA using the current state of the experimental workflow.

- Chas

Andy Fingerhut

unread,
Feb 24, 2012, 3:10:13 PM2/24/12
to cloju...@googlegroups.com
Thanks, Chas.  I updated my easier-to-follow figure to match, basically just by saying "Can go from any state directly to Screened or Accepted" in the bottom right corner.

[1] http://dev.clojure.org/display/design/Clojure+workflow+%28experimental%29

I noticed that the "Ready for Screening" -> "Screened" and "Screened" -> "Accepted" arcs are not shown as lines in your auto-generated figure.  I suppose those links are not drawn because they are covered by the "Global Actions" mentioned on the right?

I'm not sure, but I think Stuart H was also hoping for similar global actions for Resolve Issue and Close Issue, too.  Such transitions exist from most places in your workflow, except not from Ready for Screening and Screened.

That is something I'd be happy to do, given time and the proper JIRA access.

Thanks,
Andy
Reply all
Reply to author
Forward
0 new messages