On Mon, Feb 11, 2013 at 11:37:31AM +0100, Reindl Harald wrote:
Am 11.02.2013 11:31, schrieb Olav Vitters:
On Sun, Feb 10, 2013 at 07:59:22PM +0000, Ian Malone wrote:
In the end, more than any usability quibbles, the best reason to give up on a project is when it refuses to listen to its end users.
The GNOME release notes over various cycles have listed loads of changes which have been made based on the things that have been learned. This happened during 2.x as well as 3.x.
Although you do not explicitly state it, it seems you were talking about GNOME. Vincent Untz phrased it much better than I ever could, but he basically pointed at the "Power Off". You can also read the release notes for loads of other changes
this is all fine
BUT why are things completly re-written and in a pre-alpha state released replacing and destroying the users workload and after that it takes years to fix all teh issues in the one or another way?
I have a totally different view.
Could you show me the bugreport about where GNOME destroyed something on a users machine?
GNOME 3 was delayed by 2 cycles. Before that we made loads of releases available for testing. The 3.0 was really stable.
this big mistakes are happening over and over and the speed these are happening is growing with each compontent instead learn from mistakes and release software after it is finished or do not make a rewrite at all
Conflicts with release early and release often and the difference between testing by 50 people and releasing it for 500.000+.
it does users not help much if 2-3 years later things starting to get useable again - why? because in the meantime someone is changing the next subsystem against a pre-alpha and years later people are proud to have fixed a lot of issues while forget that they all were introduced by release unready software
That was addressed by Vincent during FOSDEM.
I mean: - real usability testing (help welcome!) I mean huge groups, non-biased, representing everyone, etc - real studies on biggest issues (help welcome!) I don't mean an internet survey, or a study where the outcome is 'do what some other OS does'. I mean something which is a followup on what Sun did ages ago. - better communication (help welcome!) Sometimes a huge difference to what is decided/planned and what news sites announce
e.g. the poweroff I wanted to see changed more quickly. It could've, but a study would've sped it up greatly. I mean a huge usability study at least every 2 years, and smaller ones after each release.
This to address the difference between: - one developer working on something - a few developers (project gets a few developers) - 50+ developers (jhbuild people) - 500+ people (tarballs/unstable packages) - 5000+? people (beta cycle - 3.x.0) - nothing - 500.000+? (distro release)
Every time the number of people increases 10-fold, you'll find more issues. Expecting that a few developers will ever release something that would be good enough for 500.000 is just unrealistic.