[Engine-devel] Questions about database changes

[Vitor de Lima]

Hi everyone,

I would like some feedback about how to create the upgrade script that creates columns for the architecture of each VM, template and cluster in the engine database.

The changes #17853 and #16700 introduce a field (and the respective Enum) used to store the target architecture of a cluster, VM or template (which currently can be either x86_64, ppc64 or undefined).
In their current state, these changes introduce a VARCHAR column storing the architecture, but now I wanted to implement an autocompleter for this field in the search backend, and it would be massively cleaner and easier to use the architecture field as an integer (since Enums that implement the Identifiable interface can use the EnumValueAutoCompleter class).

Considering that these two changes are already in review, should I modify them directly to use an integer or should I create another patch that changes the column in the database? If I create another patch, should it modify the upgrade script from change #16700 or it must create another script that migrates the column from a VARCHAR to an INTEGER?

Thanks,
Vitor de Lima

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/devel/attachments/20130827/70953516/attachment-0001.html>