[OSM-talk] Mechanically Cleaning Up FIXME Tags

moltonel 3x Combo moltonel at gmail.com
Wed Feb 25 12:42:09 UTC 2015


On 25/02/2015, sly (sylvain letuffe) <liste2 at letuffe.org> wrote:
> I do also agree with Frederic, imports of external data not conflated added
> with some "fixme=please fix my bad import by surveying it on the ground"
> should be remove alltogether. Good integration should be done at import time
> and should'nt rely on others by spamming the fixmes.

An import that tagged its object with fixmes is arguably a bad quality
import. But mass-removing that import's fixmes (assuming those fixmes
carry usefull info) degrades data quality further. Two bads don't make
a good.

If you're talking about fixmes that don't carry usefull info, then it
doesn't matter wether they were created by imports or not.

> For all the other manually entered fixme, a lot of care and discussion
> should be performed before attempting anything automatically.
> Even the rather useless "fixme=yes" should be handeled with care. The mapper
> might have added information that he knows to be perfectible and wanted to
> enlight that.

My guess is that manually-entered fixmes will not lend themselves to
automated fixing at all. You'd spend more time fixing your script's
heuristics than you'd spend manually doing the fixes.



More information about the talk mailing list