[OSM-talk] SteveC should decide
gerv-gmane at gerv.net
Fri Oct 2 22:02:28 BST 2009
On 01/10/09 10:40, Frederik Ramm wrote:
> If we have open issues in the community that we cannot find a good
> solution to, then the reason for this is not that we simply lack a good
> Führer who tells us what is right and what is wrong;
I may be entering dangerous waters here, but I'm wondering if this
comment of yours reveals quite a lot.
You need to distinguish between good leadership and bad leadership. Good
leadership sometimes tells people to do things they don't agree with.
Calling all such leadership Nazi is not productive. I'm now wondering
whether it's a coincidence that you work for yourself rather than for
someone else :-)
Good leadership is not the same as "makes decisions Frederik agrees
with". Good leadership is not the same as "only making decisions which
are easy because everyone agrees". Good leadership is leadership which
furthers the mission of the organization.
If Steve said that "green" and "blue" were the correct database values
for "true" and "false", then I'd write five lines of translation code
for JOSM which meant it showed up in the JOSM UI as "true" and "false",
and then I would be very thankful that the decision had been made and
all the automated scripts I'd written to use OSM data didn't have to
check for 21 different values of "false" any more. And I'd accept his
decision on the basis that any decision, in this case, is better than no
decision, and that I trust him to have good reasons for making it the
way he did.
Having said that, I am also pretty sure he wouldn't say that.
> it is because these
> issues are difficult and the community is perhaps divided about it. We
> do not need anybody to make a decision in these cases; that doesn't help
> at all.
No, it's precisely what we need.
> things simpler. It seems that you would prefer a wrong decision over no
> decision at all - but why do we need decisions at all? If there are
> issues where the community cannot make up their mind, can you not just
> live with that and arrange your technology in a way to deal with that?
Because that way lies misery and code complexity.
More examples from the Mozilla project: if one vocal group want
something one way, and another vocal group want something the other way
in Firefox, the _worst_ thing you can do is make it a preference so that
both sides can have what they want. That just makes everyone's life more
difficult, because there are now two code paths to test and maintain.
Multiply this up by a number of decisions and you get complexity explosion.
If I were considering using OSM data in my business, I would consider it
laughable that after 5 years there had not yet been a decision on what
value or small set of values I needed to look for on boolean attributes
to see whether they were true or false. Laughable.
More information about the talk