Re: Normalising UDD to some extend?
On 19/02/09 at 13:09 +0100, Andreas Tille wrote:
> On Thu, 19 Feb 2009, Lucas Nussbaum wrote:
>
>> Well, Depends: can obviously change between architectures. And there's
>> no reason why other fields wouldn't be different as well, since control
>> files are generated on a per-arch basis during the build process.
>
> OK, I'm definitely convinced in this aspect. ;-)
> The main point I was worrying about were actually description and
> long_description which make a big part of the data. Do you think
> it makes sense to try some normalisation here (because I fail to
> see a reason why it should be arch dependant)?
Having a big single table might be a performance win compared to several
shorter tables that you would often need to join.
Also, during import, for each package, you would need to compare if the
already stored description matches the one being inserted.
Btw, the following packages have different descriptions or long
descriptions on different arches:
default-jdk
default-jre
default-jre-headless
gedit-plugins
libxenomai1
nvidia-kernel-2.6.26-1-amd64
openjdk-6-jre
openjdk-6-jre-headless
qvwm (only alpha different from others, apparently)
--
| Lucas Nussbaum
| lucas@lucas-nussbaum.net http://www.lucas-nussbaum.net/ |
| jabber: lucas@nussbaum.fr GPG: 1024D/023B3F4F |
Reply to: