Since the data files and code are likely to be reworked in the process of 
pushing this new project forward, I would recommend we take some time to review 
the structure of the DDR and its codebases. In writing a C client (dClass), I 
found some areas of improvements:

-I found the need of builders unnecessary. The approach I took in dClass was to 
classify the pattern by a type: strong, chained, and weak (there is also a way 
to break ties when multiple patterns match). These can be adjusted if needed, 
but I found that this is was fastest and most flexible way to match all 
patterns against a user-agent string.

-In doing the above, I was able to extend the DDR to detect browsers, operating 
systems, and bots.

-I was able to achieve a high level of performance speedup in using a hybrid 
search/pattern data structure as the core of my client: dtree.

I just want to get the ball rolling and possibly start a technical discussion 
as to the code structure of this project.

My client is here for reference: https://github.com/TheWeatherChannel/dClass

Thanks,
Reza

-----Original Message-----
From: Bertrand Delacretaz [mailto:[email protected]] 
Sent: Tuesday, October 09, 2012 8:41 AM
To: [email protected]
Subject: Re: Status of openddr-resources xml files ?

Hi,

On Mon, Oct 8, 2012 at 7:39 PM, Laurent Perez <[email protected]> wrote:
> ...Are there plans to donate these resources to the Apache foundation ?...

OpenDDR actually intended to do so, but after discussing it the Apache 
Incubator team preferred to accept only the code so far, details are at 
https://issues.apache.org/jira/browse/DMAP-11

We're starting to discuss how to go forward, it seems to me like the best 
option will be to setup DeviceMap to accept data contributions in a similar way 
than OpenDDR does, for new data, and also adapt the data parsers to be able to 
use other sources so that users can get a complete database.

-Bertrand

Reply via email to