[opensource-dev] from debug to prefences
Oz Linden (Scott Lawrence)
oz at lindenlab.com
Tue Jan 25 06:35:13 PST 2011
On 2011-01-24 20:45, Robert Martin wrote:
> On Mon, Jan 24, 2011 at 7:53 PM, Trilo Byte<trilobyte550m at gmail.com> wrote:
>> Perhaps the experimental/developer-related debug settings should be accessible from the Developer menu (further removing it from casual users). Then an 'Advanced Preferences' floater could be made available from the Advanced Menu. This would allow for an effort to seriously streamline the main/basic SL preferences.
> what needs to happen is a full bore IRS level accounting needs to be
> done on which preferences are actually used and which are
> outdated 2.* series prefs (pull these out)
> outdated 1.* series prefs (pull these out)
> oddball but still used prefs
> dangerous prefs
> normal and menu accessable prefs
> power user prefs
> nonlinden admin prefs
> Linden Only Prefs
> TPV prefs that slipped into the stock viewer (was the feature added to stock??)
>
> and then sit down and document what each one of those does (and
> safe/dangerous values for same
Which illustrates nicely the scope of the problem, and explains part of
why adding options is costly down the road.
The plain fact is that usually the way we discover which preferences are
used is to remove (or relocate) them and see who screams (we do get
prefs settings as part of crash reports, but that only tells us what
people who crash are using). As far as we can tell, someone always
does, and correctly or not thinks that we've hopelessly broken something
critical.
The other problem is that some settings become "folklore": users, for
whatever reason, become convinced that if you set preference X to Y then
bad thing Z doesn't happen (or good thing Z does) when in fact there is
no relationship at all (or, in fairness, we didn't think that there was
and the relationship itself is a bug).
So... creating the categorized list you suggest is itself a very large
job, the results of which will be at best controversial and more
probably just wrong. That doesn't mean it's not a noble undertaking
(the most heroic quests are sometimes futile), but it's a tough sell.
More information about the opensource-dev
mailing list