On Mon, Jun 7, 2010 at 3:07 AM, Conscious User
<conscioususer@xxxxxxx> wrote:
Hey all,
I want to finally write something I've meant to write for a long time.
First off, here is some context, in the form of recent events:
Sense Hofstede mentions that NotifyOSD has barely been touched in
the latest Ubuntu iterations:
http://twitter.com/qense/status/14779713532
Matthew Paul Thomas expresses dissatisfaction with the current
shape of NotifyOSD:
https://twitter.com/mpt/statuses/15525108146
https://twitter.com/mpt/statuses/15564045224
https://twitter.com/conscioususer/statuses/15564426984
https://twitter.com/mpt/statuses/15584217252
Benjamin Humphrey posts a long essay about the current unpolished
elements in Ubuntu. I don't agree with the entire post, but I
do agree with its essence:
http://www.omgubuntu.co.uk/2010/06/many-hands-make-light-work-few-make-it.html
David Siegel, in a comment to this posts, explicitly says:
"A good first step would be to decide not to ship features if
they're not finished yet!"
It seems that ever since Ayatana started giving fruit, new ideas
have been proposed and started faster than old ideas are being
polished enough. Perhaps "polished" is not even the word, is
at implies improvement over something already complete. And
some things are not yet complete.
For example, NotifyOSD, which IIRC was the first ever Ayatana
implementation to land, is still not fully following its
specification. For example, the morphing windows/alerts do
not exist yet, and they would be helpful in improving the
current update-manager behavior (one of the hottest bugs ever).
Dynamic timeouts based on text length do not exist yet, as well.
https://bugs.launchpad.net/ubuntu/+source/update-notifier/+bug/332945
Another example is the Me Menu, which created a lot of confusion
with respect to what the text field was supposed to do, confusion
that might not exist that the specification, which included
microblogging icons, was being fully followed.
I don't think that there is a single Ayatana project which
already reached a stated we can call close to "finished".
To make things worse, the user is expected to figure out those
incomplete implementations by himself. The Manual Project was
an improvement in this regard, but that still doesn't change
the fact that the new user is being faced with incomplete
implementations that might be very intuitive in the future
but currently are not.
When I suggested some kind of "getting started" tutorial, to
complement the install slideshow, I was shot down in this list
(twice) with the argument that "it's better if the features
are intuitive and speak for themselves". Well, while the
implementation is incomplete, that won't happen. What
should be done meanwhile?
Meanwhile, Maverick is proposing Window Indicators, which will
require a lot of technical work and upstream collaboration.
And probably will take a lot of time from the few people who
are responsible for the majority of NotifyOSD/indicator
coding. Time that, perhaps, could be better spent in fulfilling
popular requests.
My question is: isn't it time to put some brakes on the
enthusiasm and start prioritizing polishing instead of new
features? The current approach is not scalable, and this is
starting to show...