[Rebuild] Getting underway before 1st April
errt at gmx.de
errt at gmx.de
Tue Jan 31 18:33:11 GMT 2012
Am 31.1.2012 17:32, schrieb Michael Collinson:
> Taking Frederik's comments into account, I would amend my support to:
> we should be working on running actual scripts ASAP, whether they be
> bboxes or off-line copies. [Caveat: I fully acknowledge that I am now
> a talker volunteering other folk's time.]
I do agree we should really getting to work on the actual scripts, but
there's nothing we can do before we finally decided on the way to go. We
should definitely work with (probably small but interesting, e.g.
extracts from Czech Republic or so) offline copies, to make sure we
don't mess up the database. Also this is the only way that allows us to
do changes to ids and changesets and such, which I still prefer.
> - (for me to continue) to persuade the community that the WTFE algo
> should be the basis of the transition
I think the algorithms of the WTFE are quite good to show what is
problematic and what not and so should be the basis of whatever we do.
But still they just mark objects as problematic or not and don't state
anything about HOW the problems will be resolved, e.g. what tags will be
left and with what values, what nodes or relation members will stay and
in what order and so on.
> - that split/merged ways is the only point of contention that could
> make any potential real difference
Yes, these are the most problematic cases, while most other situations
have been discussed quite extensively before, we still have no
statements about how to deal with splits and merges. We might even
consider just ignoring them if that's ok from a legal point of view, but
if we can, we should try to reconstruct them. This will probably make
changes to prior changesets necessary, I think.
> - that the WTFE "rules" could indeed by mapped into scripts that
> actually altered the database.
As stated above, I don't think this is possible. These rules give good
hints about what to consider when creating the scripts, but they still
are only presenting problems, not solving them.
> - and that the resulting map would look almost exactly the same as the
> CleanMap Simon is publishing now?
That's not very likely. While the CleanMap is a great tool to find
problematic objects and areas, it is quite simplified and probably far
from what will be the real result of the changeover.
> Or am I smoking something I shouldn't?
Let me have a toke and try :p
More information about the Rebuild