[OSM-legal-talk] Licenses for Produced Works under ODbL

[Igor Brejc]

Hi Eugene,

Thanks for your thoughts. My comments are inside.

On Mon, Oct 29, 2012 at 11:49 PM, Eugene Alvin Villar <seav80 at gmail.com>wrote:

>
> A rendering database is a database that is used to render/draw a map.
> Raw OSM data is not often suitable for rendering maps and you need to
> preprocess OSM data into an intermediate database like the PostGIS DB
> produced by Osm2pgsql.
>

Right. But the problem is this: between that rendering database and the
final product there are also some other in-memory data structures which
renderers use to actually draw stuff. One example I can give is creating a
highway network graph in memory and then searching for paths with same
street names so the label placement can be optimized. OK, if you use
PostGIS then this may be done directly using SQL queries, but what happens
if you don't? Maperitive, for example, runs everything in-memory from an
OSM XML file.

Some then say that these in-memory data structures are also Derivative
Databases. In what form can you then offer such a Database to someone that
requests it?


>
> Take note that releasing an algorithm is just an alternate for
> releasing the derivative ODbL database. And from the wording of the
> ODbL, yes, the algorithm doesn't have to be source code, just
> "machine-readable" which can mean any electronic text like: "Use the
> program Osm2pgsql with the following settings on the following OSM
> extract..."
>

One of the difficulties with discussions like these is that they tend to
use open-source software as examples :). That misses the point. What
happens if the electronic text says:

"Use the program ArcGIS with the following settings on the following
OSM extract...", or indeed "I used my own unreleased program on the
following OSM extract..."?

If this is valid case for ODbL, then the whole "method" clause becomes
nonsensical. I can understand the intentions of people who wrote ODbL. But
the problem is that ODbL IMHO is poorly designed and opens up a whole lot
of issues to which you can have very different answers and this doesn't
help anyone.


> And if you want to release source code, it can be under any license
> with a reasonable cost or free if over the Internet. There is no
> obligation for the recipient to share with others. You can actually
> say, "here's the source code, but you are not allowed to share it with
> others."
>

OK, but I see several problems with this situation:

   1. ODbL doesn't really say anything on the matter of licensing that
   machine-readable stuff. So it's open to interpretations.
   2. Not one company will dare to give out their proprietary source code
   to someone, even if they release it under a very strict license. The risks
   of someone inadvertently then pasting that code on pastebin (example) are
   just too great - and there's no way back.
   3. How much source code is enough? Is there a requirement that it should
   compile and that the receiver should be able to repeat the process of DB
   derivation?
   4. What's the purpose of it all, anyway? :) If someone releases the
   source code to a single person which then cannot share it with others, how
   does the larger OSM community then benefit from it all?

Again, I really don't see how the 4.6 clause can be applied in some
meaningful way, other than scaring "small players" off and leaving the
ground to big guys with multi-billion budgets who aren't really afraid of
ODbL anyway. That's why we need some well-thought-through community
guidelines which address closed source software, too.

Best regards,
Igor
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/legal-talk/attachments/20121030/b7189315/attachment-0001.html>