incubator-bloodhound-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Olemis Lang <ole...@gmail.com>
Subject Re: Datamodel and data consistency
Date Thu, 25 Oct 2012 03:52:26 GMT
On 10/24/12, Peter Koželj <peter@digiverse.si> wrote:
> Jure and I have been looking at the datamodel to see what would it take to
> implement some of the enhancements that we have in mind.
> Here are a couple of observations that pop up:
>
> 1. Parent/child relationships are not model by permanent ids but by names
> (like version, milestone, product... name) that can be changed by user
> through admin interface. A bit unusual but simple and eliminates the need
> for joins when no additional data from the other entity is needed.
>

This is what Trac-dev should be working towards regarding relations of
any kind between resources .

http://trac.edgewall.org/wiki/TracDev/Proposals/TracRelations

> On the other side, renames of versions, milestones, products... (which are
> indeed rare) require update of related tickets.
> I imagine that renaming a product which has tens of thousands of tickets
> could be a bit problematic. Much more worrying is that at least some of
> these renames update tickets individually instead of as a single
> SQL statement. At least this should be fixed.
>

I'm not sure of whether I understood correctly , but transactions
might help to roll back unwanted modifications ...

> 2. There is no referential integrity imposed on the db level. But what is
> even worse is that there is no data integrity check on application level
> either.

Your experiment yields accurate conclusions . Maybe there will be some
answers if you forward your concerns to trac-dev .

[...]

-- 
Regards,

Olemis.

Blog ES: http://simelo-es.blogspot.com/
Blog EN: http://simelo-en.blogspot.com/

Featured article:

Mime
View raw message