Multiple Executions Causes Kanban Board Issues about tfsaggregator

[giuliov]

First of all thanks: you are a brave beta tester!
Honestly I have not tested the Board scenario, but I suggest considering reentrancy: after you touch the work item and the plugin saves it, TFS will trigger another change notification that reenters the plugin for the same work item.

[shaneedwards]

Thanks for the response, hope you had a good weekend.

I wanted to follow up with more information after additional testing and see if you could offer any further advice. I was able to find a pattern--this occurs any time I operate on the user story via TFSAggregator using C# scripting. I do see the re-entrancy that you mentioned which I understand, based on some comments in other forums, is intentional. I have handled that problem by setting a 'last modified' field and checking in an IF so that when I make a change it will not process more than once. The problem I am seeing is that even when I do nothing during that second operation it appears that TFS gets into some state where the next transition will always fail. It turns out, this isn't just a Kanban issue, but happens if I manually transition via the form as well.

I will post my slightly modified code and a video of the interaction as well. In the code snapshot below I have highlighted the two debug outputs that you'll see in the trace window.

In this video you'll see me transition to a state that does not run the C# aggregation code (falls through to the else), then transition to the state that runs the main body of the code and you'll see in the debug that the first run works fine and the second (reentry) falls through to the else. Then, the third transition just silently fails. No debug output, no errors on the screen, just flat rejected.

https://www.youtube.com/watch?v=MEm5f8Lc9pM

Any insight you can give on what might be cause the issue would be of great help!

[jessehouwing]

This may be related to the fact that the web ui doesn't expect the
underlying work item to be changed.

TFS 2015 will make the front-end aware of backend changes through signalR
I've heard, which would solve this issue. I don't think there is anything
you can do to prevent this, other than reloading the screen after a change
if you intend to update the same WorkItem multiple times.
On Jun 15, 2015 4:30 PM, "shaneedwards" [email protected] wrote:

Thanks for the response, hope you had a good weekend.

I wanted to follow up with more information after additional testing and
see if you could offer any further advice. I was able to find a
pattern--this occurs any time I operate on the user story via
TFSAggregator. I do see the reentrancy that you mentioned which, based on
some comments in other forums is intentional. I have handled that problem
by setting a 'last modified' field and checking in an IF so that when I
make a change it will not process more than once. The problem I am seeing
is that even when I do nothing it appears that TFS gets into some state
where the next transition will always fail. It turns out, this isn't just a
Kanban issue, but happens if I manually transition as well.

I will post my slightly modified code and a video of the interaction as
well. In the code snapshot below I have highlighted the two debug outputs
that you'll see in the trace window.

[image: 2015-06-15_09-24-08]
https://cloud.githubusercontent.com/assets/7059852/8162042/80de38a6-1340-11e5-8a0f-c22849b8c19a.png

In this video you'll see me transition to a state that does not run the C#
aggregation code (falls through to the else), then transition to the state
that runs the code and you'll see in the debug that the first run works
fine and the second (reentry) falls through to the else. Then, the third
transition just silently fails. No debug output, no errors on the screen,
just flat rejected.

https://www.youtube.com/watch?v=MEm5f8Lc9pM

Any insight you can give on what might be cause the issue would be of
great help!

—
Reply to this email directly or view it on GitHub
#36 (comment)
.

[shaneedwards]

Ahhh, that makes sense. Explains why I wasn't finding an error in the back end or in JavaScript. It isn't an error, just the UI getting out of sync. Not to belabor the point, but is there a way to fire off a reload event on the front-end from TFSAggregator?

[jessehouwing]

Unfortunately not. That will only happen with the signalR support.
On Jun 15, 2015 7:43 PM, "Shane Edwards" [email protected] wrote:

Ahhh, that makes sense. Explains why I wasn't finding an error in the back
end or in JavaScript. It isn't an error, just the UI getting out of sync.
Not to belabor the point, but is there a way to fire off a reload event on
the front-end from TFSAggregator?

—
Reply to this email directly or view it on GitHub
#36 (comment)
.

[shaneedwards]

Thank you for the assistance. I'll be waiting patiently for 2015!