Re: [ ACTION-107] Updating version list identifier algorithm

On 9/19/07, Mark Baker <distobj@acm.org> wrote:
> On 9/18/07, Ian Hickson <ian@hixie.ch> wrote:
> > In any case, I would recommend not parsing versons at all. Just treat them
> > as opaque strings, and assume that there is an upgrade if the version on
> > the server doesn't match the version in the client.
>
> Sounds good to me, but then you have to wonder why etags wouldn't suffice.
>

I guess because you don't always have an etag to check against: like
if you send me a widget in an email or over BlueTooth. And because
widgets may be sometimes automatically generated (eg. Opera's
Widgetizer [1]), but the version stays the same for every widget even
though some of the (auto-generated) internals may be different. Also,
you might make an update to a widget, like making an image slightly
brighter,  but without updating the version number or forcing everyone
to update based on such a minor aesthetic change. Making such a change
would change the etag on the server, but does not warrant a download.

I think etags are probably part of the solution, but not on their own.

Kind regards,
[1] http://widgets.opera.com/widgetize/

-- 
Marcos Caceres
http://datadriven.com.au