On 19. Jun, 2013, at 20:13, Gary Martin wrote:

> On 19/06/13 09:54, Branko Čibej wrote:
>> On 19.06.2013 10:05, Andrej Golcov wrote:
>>>> Yes, that's the straightforward reasoning. Just notice that workflow
>>>> actions not necessarily have to modify ticket status. The most common
>>>> example is leave action.
>>> I mean: deleting or adding relation does not change ticket at all, IOW
>>> this is not supposed to be ticket.save_changes action. Currently, user
>>> is required to have *_CHANGE permission on resource (e.g.) to change
>>> its relation but this can be easily changed in future. I can imagine
>>> the business requirement that user can make relation between tickets
>>> that he/she doesn't have change permission.
>> I'd suggest that adding a duplicate relation does not necessarily imply
>> that one of the tickets involved should be "resolved as duplicate". It
>> is a valid, although unusual, workflow to mark as duplicate (possibly of
>> several other tickets), and yet leave the ticket state unchanged.
>> 
>> So I propose that the duplicate relation itself should not be special;
>> rather, the "resove as duplicate" action should take account of existing
>> duplciate relations, possibly still allowing the user to create yet
>> another such relation.
>> 
>> -- Brane
>> 
> 
> That is twice I agree with Brane completely in relatively quick succession!
> 
> I have no problem with users marking tickets as duplicating outside of the 
> resolving tickets as duplicate method. Under those circumstances, it would 
> not be good to remove the relationship automatically on reopening as we 
> couldn't second-guess the reason for the relationship being added.
> 

Ok, I suppose that's the way to proceed then. Am I correct to assume that
resolving the ticket as duplicate should still require the user to enter
a duplicate ID, as mentioned in the BEP? The current (Trac) implementation
of duplicates without references is a bit awkward in that regard.

> Cheers,
>    Gary

Reply via email to