tom at compton.nu
Tue Dec 20 17:48:45 UTC 2011
On 20/12/11 14:18, Kate Chapman wrote:
> On Tue, Dec 20, 2011 at 8:01 PM, Tom Hughes<tom at compton.nu> wrote:
>> I would be very reluctant personally to have a paid implementation because
>> it basically pays somebody to the interesting work of writing something
>> while leaving the grunt work of trying to review the results to the
> I'm curious Tom, is implementing every feature interesting to you?
> Are there tasks other than reviewing patches that you consider grunt
Of course I was generalising with the comment about writing features
being the interesting bit.
My key point really is that reviewing and integrating things is a
bottleneck in our system and just contracting people to feed more and
more code of possibly dubious quality into one end of that bottleneck is
not likely to improve that.
I freely admit that this is partly a problem of my making and that
probably I need to find better ways to handle code reviews as I am far
too prone to fixing stuff myself rather than sending it back with
comments for the contributor to fix issues.
It's partly that much of the time it would take me just as long to
describe the issue as to fix it, though that ignore the longer term
advantage of hopefully teaching the contributor how to do things better
so that next time less changes are needed.
Maybe there are better tools we could use. I know there are web based
code review tools and things, though I have never used any. Maybe we
should be looking at some of those.
> I personally think you pay people to do the work people don't want to
> do. But I've never been 100% clear on what that is in OSM. Though my
> assumption really is most of the time now it is people having time to
> do the work they want.
The problem with paying people to do the work that people don't want to
do of course, is that the people you're paying probably won't be that
thrilled by it either and are therefore not likely to do a good job.
Now that may be fine for some tasks, where something is either done or
it isn't, but for an intellectual activity like programming there are a
wide range of values for "done" with results that vary greatly in their
What I am afraid of is a lowest price bidder turning up one day with a
metric tonne of hastily knocked out code and the community expecting
that to be something that can just be rolled out. Perhaps even more
worrying is the amount of work I might have to do to make such code usable.
Tom Hughes (tom at compton.nu)
More information about the osmf-talk