launchpad-dev team mailing list archive

[Jonathan Lange <jml@xxxxxxxxxxxxx>]

On Fri, Jan 29, 2010 at 12:22 PM, Abel Deuring
<abel.deuring@xxxxxxxxxxxxx> wrote:
> On 29.01.2010 10:54, Jonathan Lange wrote:
>> Sorry, I had meant to reply to all.
>>
>> On Fri, Jan 29, 2010 at 9:53 AM, Jonathan Lange <jml@xxxxxxxxxxxxx> wrote:
>>> On Thu, Jan 28, 2010 at 7:30 PM, Abel Deuring
>>> <abel.deuring@xxxxxxxxxxxxx> wrote:
...
>>>> In other words, a deliberate denormalisation. What do you think, should
>>>> we start with a plain query without such a denormalsation, or do you
>>>> think that it is better to add the "cache column"
>>>> date_created_youngest_patch to the Bug table?
>>>>
>>> This is really Stuart's province. I don't have a strong opinion on this.
>>>
>>> However, it seems to me that we're getting increasingly frequent
>>> requests to denormalize data in order to make querying faster. Perhaps
>>> we need a general answer, or some other technology.
>
> Perhaps, but finding a general answer is not trivial, I think. "Think
> harder to find a better query" is one answer, but won't give useful
> results in each case; "We don't really need such a query" is another
> (though Bryce had good reasons to ask for the sort order we're
> discussing); "Let's accept denormalisation after thoughtful
> consideration if we find no better way" seems most reasonable to me.
>
> Regarding other technology, I'm quite curious about suggestions.
>

Mostly I mean a data warehouse.

In any case, this is Stuart's call. It's about the database and not
really about the nature of the product.

jml