[OSM-talk] Server slowness

[Richard Fairhurst]

Martin Spott <Martin.Spott at mgras.net> wrote:

> SteveC wrote:
>> Sigh. We're building a wiki, not a 1970's GIS. Most of them just don't
>> apply. We have fundamentally different problems. Nobody cares what
>> format the nodes are in right now.
>
> I'm sorry that the discussuion lead to some irritation at your end:
> Nobody's talking about 1970's GIS.

But you do need to recognise quite how alien "GIS", as a discipline of  
any age, can look to outsiders.

Quite how it happened I'm unsure, but to Joe Hacker, GIS can seem a  
very insular science: cut off both from the mainstream of back-end  
software development and the popular front-end of good cartography.  
ArcGIS, to take the market leader, looks about as appealing to an  
agile developer as SAP does (and has about as convincing a UI). GRASS,  
the best-known open source GIS, can doubtless do cool stuff if you're  
into it but, honestly, to the new user it _does_ look like a 70s  
product. Nor is the image of GIS helped by so many public-facing  
web-GIS implementations, like those webmaps you see on local council  
websites with 200 molasses-like layers. You can see how the perception  
develops.

OSM is, at present, a typical open source community of developers,  
hackers, idealists and agitators (and ninjas, don't forget the ninjas:  
http://www.charlbury.org/ninjas). There is a very steep learning curve  
to traditional GIS, and open source developers tend to learn by  
osmosis rather than reading interminable docs (I like "a slower death  
than being trampled by chickens" - nicely summed up). GIS is never  
going to catch on in the open source community while that's true.

To most of us, I suspect, GIS technology seems like a really, really  
complex way of solving a not-that-unusual DB problem.

At present its two most compelling advantages for OSM seem to be an  
easier import of US data and a possible performance increase. The  
former would be good, but there's never really been any clamour for US  
coverage in OSM, so I guess that's not going to convince people. The  
latter - well, wouldn't we all like a faster server? But as  
developers, we'd also like OSM to be as easy to install and hack on  
our home machines as it is now. I'm not sure that a 40% speed increase  
vs. confusing all our existing developers is worth it.

cheers
Richard