[Loadstone] OSM converters and enumerate points.

Lex lex at progger.ru
Sun Oct 11 07:29:14 BST 2009


Hi Rob and all,

First of all, thanks for your testing. It is greatly appreciated.

11.10.2009 8:34, Rob Melchers пишет:
> Now that the Python converter also works here I took the chance to 
> compare this tool with the converter as offered by the off-line tools. 
> As a test file I used the wales .osm file (134 MB) to get a good 
> comparison for speed. 
In deed, python converter is not ready for such large files. Method i am 
using requires a lot of memory but gives very speedy results on files 
lesser than 50 mb. In my practice such size is quite enough. If you 
really need to convert such large entities, i will think about how to 
adopt my tool. Anyway, can you please test both converters on file under 
50 megabytes?
> However, the PHP converter gives a result file with 33020 points, the 
> Python result only contains 17530 points. It seems to me a lot of data 
> loss against 40% speed gain. Unless the Python tool improves in point 
> extraction I rather wait 3 minutes more. 
I am afraid python converter did not finish its work. can you provide me 
a log? In other case, i'd like to debug that situation because with my 
tests python converter extracts more points (for example, adresses where 
street information is absent, only house number) etc.
> While testing I discovered that both tools do not restrict the name 
> field to 75 chars, this has been fixed in the PHP version in the 
> off-line tools. 
I don't want to have such restriction by default. may be adding it as an 
option will be a good solution. In my opinion it is more appropriate to 
fix strings manually in most situations.
> Also updated is the database reprocessor, it now has the ability to 
> enumerate points with equal names according to position. 
I don't fully understand what do you mean. can you give more details 
about this new feature?



More information about the Loadstone mailing list