-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Diego Moya wrote on 21/09/10 16:48: > > On 21 September 2010 15:57, Matthew Paul Thomas wrote: >... >> For the same reason operating systems don't have visible controls for >> configuring the appearance of tooltips. If they're designed right, >> they should be below the threshold of triviality. If a substantial >> proportion of users want to configure them, there's something wrong >> that we need to fix. > > Both situations are not really comparable. Notifications have proven to > be a more difficult problem than tooltips, which are basically local to > a single widget each time (and notifications are unrelated to the > content below them). Notifications are global to the system, so you > should ultimately take the whole system into account to get it right.
Perhaps scrollbars are a better example, then. Scrollbars as we know them are deceptively complex, and took about three years to get right <http://folklore.org/StoryView.py?story=Busy_Being_Born.txt>. But pretty much the only configuration you can make to them now is changing their width in Windows, and their arrow button positions in Mac OS X. > The smallest misappreciation could produce a less-than-optimal > solution. You should be omniscient to create your "right design"! Not omniscient, just following a standard design process. > That scorched earth approach of zero-configuration ("if it ain't > perfect, disable it!") means that until you get it right some people > will suffer in the meantime without a way to fix it, even assuming that > the perfect design exists. There's nothing "disabled" as far as I know. But every added option would make it harder to implement and to test. > Also think that "configuration" doesn't need to be "panel with check > boxes". Allowing direct manipulation of bubbles (for example to > reposition them) would potentially fix with a one-time drag-and-drop > any problem of obscured content, without being a burden to the user. >... > This is a personal preference, but I really don't like the idea that > clicking on a widget hidden below the notification should have some > effect. The whole "click-transparent notification" idea in the current > design is disturbing to me, it always felt weird and uncertain on what > would be the result. Maybe, then, it would make sense for the bubble to bob out of the way (e.g. down a bit) when you go to the corner, and stay there. No clicking required. >>> - Don't show notifications if there has been user interaction in the >>> previous 5 seconds. >... > In my suggestions I'm relying heavily on the notion that missing a > notification is not a big deal - I think that was one of the design > guidelines. I'm just pushing it to the extreme in order to reduce > interruptions to the user flow while working with application content. > >>> - Even if the bubble is closed, allow the user to reopen it from >>> the Me menu. >> >> Many if not most notification bubbles have nothing to do with social >> networks or instant messaging, so that would be a poor fit. > > ... or somewhere else, where logging that class of notifications was > appropriate. If there isn't a sensible place where to log the message > in a particular bubble, it wasn't that important to begin with. These points seem to contradict each other a bit. Our current approach is that yes, missing a notification is not a big deal, and *therefore* it's not important enough to have a global notification log. One reason it might not be a big deal is because there is an application-specific log, e.g. an IM chat. - -- Matthew Paul Thomas http://mpt.net.nz/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAkykmS0ACgkQ6PUxNfU6ecqlgwCghxfvj+l/oR2Bpo7NEv8M+P2I FIcAoJKfR6fy9MuOoUzho7Bjx74EMEj2 =uoVA -----END PGP SIGNATURE----- _______________________________________________ Mailing list: https://launchpad.net/~ayatana Post to : [email protected] Unsubscribe : https://launchpad.net/~ayatana More help : https://help.launchpad.net/ListHelp
