On Dec 6, 2007 9:28 AM, David Turner <
novalis@openplans.org> wrote:
On Wed, 2007-12-05 at 20:23 -0500, whit wrote:
> we might consider better ids to identify projects and people to
> communicate between apps.
>
> both projects and people already have uids associated with them, it
> would avoid certain problems by using them. Currently I'm thinking of
> project deletion or incomplete project creation where lingering
> instances of the id create issue when someone tries to use that id
> again.
Do you mean a random number? If so, I'm -1 on that.
Random numbers are really nice inside a monolithic system, because you
can always resolve them back to what they refer to. But in a
distributed system, it can be a hassle to figure out why you're seeing
152352-134214-412 when you expect 912334-613412-678. I would rather see
URLs as ids, since they're still unique, but they're easy to understand.
urls are not unique in time (ala the scenario I suggested initially). iow, they can be filled by in.
readability definitely does matter, but for CRUD operations, I would put avoiding collisions over readable ids.
-w
--
Archive:
http://www.openplans.org/projects/opencore/lists/opencore-dev/archive/2007/12/1196954846229