[Legal-general] question re. OSMF's attitude to hosting alternative licence servers

TimSC mapping at sheerman-chase.org.uk
Thu Jul 29 14:14:22 BST 2010


On 29/07/10 11:20, Iván Sánchez Ortega wrote:
> OSMF's resources are limited. Both money and sysadmin time. If you have a look
> at the DB and rendering server specs, you'll see they are high-end machines,
> and that 30 gigs of ram and 15 SAS drives don't come cheap. Not to speak
> about decent sysadmins, which are very hard to come by.
>    
Thanks for responding.

The point about the lack of sysadmins is a good one. OSM has never lost 
one of my edits - I fact I find more amazing as time passes! As far as I 
can tell, this lack of sysadmins is the primary reason why a PD server 
has not taken off yet. I was thinking about putting some effort in 
myself. I do have the rails port running and I had made software 
changes, but I am no expert.

But I don't see your point about hardware resources carrying much water. 
We don't know what the hardware requirements are going to be at this 
stage. I imagine, given the low traffic on this list, only little 
resources are needed at least to begin with. The comparison with the 
main server hardware spec is simply not valid. Given that OSMF has 
relatively healthy(?) accounts and the ease the foundation met their 
previous appeals for donations towards servers, are you seriously 
suggesting that OSMF don't have the hardware/financial resources? Or is 
it you are unwilling to commit those resources?

To clarify what I mean by "hosting an alternatively licensed server", I 
should add: a relatively modest server would be needed for beta testing 
(involving 10s not 10,000s of users) and software development (to modify 
the rails port to a usable state for a fork). After that, OSMF keeps the 
situation under review and takes action depending on user demand and 
available resources. But that plan would need support from the current 
sysadmins, I am just outlining a possible scenario. What I am not 
suggesting is to duplicate the  main database server infrastructure!

> The real problem is not OSMF resources, but the ability to grab the attention
> of potential contributors.
You seem to be implying that we need a significant number of users to 
make this worth while. I don't see that necessarily follows. Having a 
few towns intensively mapped by a few hard core users and available as 
PD would be a worthwhile achievement.

> Now, supporting two datasets with different licenses, and having to explain
> why to potential contributors? That's just nuts.
>    
This is actually the main objection I expected from OSMF. There is a 
real risk it would make things more confusing for users, if we handle 
this poorly. But I am not proposing "equal billing" for an SA and PD 
license. Or even equal support. We should not expect contributors to be 
involved with both, unless they really want to (like me). Therefore the 
effort to support them is not doubled. We should present information to 
new users at a rate when they are ready to take it in. I foresee at 
least a few months of beta testing a new server with a small core of 
users. This itself won't even be visible to new users.

There may even be some contributors who would only contribute to a PD 
dataset. This would increase the pool of contributors beyond what can be 
achieved by PD. And I am not exactly a light weight mapping contributor 
myself!

> The effort required for
> doing an import would easily go double or triple, just because of explaining
> the licenses and writing longer agreements with mapping agencies.
Then the simplest thing is to not import data from these sources. A one 
size fits all approach is not necessary. I don't think we should aim to 
have the same data in each server!

I note that the current contributor terms effectively enables the data 
to be released as PD (given a vote, etc). So really the negotiation is 
pretty similar for the CT and PD, unless the CT is revised, which is a 
possibility.
>> Since the purpose of OSMF is to support the mapping community, [...]
>>      
> No, it's not.
>    
Yes, I think you'll find it is. To quote the front page of the 
foundation's wiki: "The OpenStreetMap Foundation is an international 
non-profit organisation supporting but not controlling the [open street 
map] project."

Given there is actually significant interest in PD, based on the doodle 
poll, I hope that OSMF actually does less "controlling" and more 
"supporting". In fact dual SA and PD is more popular than exclusively 
SA. It is quite possible that PD is more popular than SA overall, but 
OSMF seems to have been careful not to ask that question of 
contributors. You have expressed your personal view that SA is the way 
forward (which is fine), but it doesn't follow that OSMF should limit 
the alternatives, particularly against the wishes of the OSM community. 
I partly blame the duration and uncertainty of the ODbL relicensing on 
this controlling attitude of OSMF.

One issue that has not been raised is what is the future of the license 
beyond ODbL v1? And does that negate the need of a PD fork, if the 
project goes down that road anyway (I feel that possibility is unlikely, 
but think some hold this view).

I am also anticipating SteveC entering this discussion. It seems that 
much of the OSM core inferstructure (the license, the database and a way 
to edit it) was begun using his own effort. I wonder if anything can be 
added to this core without his support. With SteveC being anti-PD, I am 
expecting a rough time...

Anyway, hopefully OSMF can do what it can to support the pro-PD 
contributors. Given that I have attempted to address all your points, 
I'd be interested in any rebuttals.

Regards,

TimSC




More information about the Legal-general mailing list