Processing Work Flow

[woodbri]

It seems to me that we need to define the processing work flow. For example:

• processing command line arguments
• reading data
  • what formats? what tools/libraries?
  • what tags
  • do we plan to use osrm lua scripting
  • how do we handle turn restrictions
• processing the data
  • what processing needs to take place
  • do we node it here or later in postgres with postgis/or pgr_nodenetwork
  • does the processing need to modify the node ids if yes how do we track the originals
• writing data to postgres
  • what are the tables that will get created: nodes, edges, restrictions, others?
  • should names be extracted into a separate table and let the edges point to that table?
  • what columns and types
• postprocessing in the database?
  • what needs to be done, why?
  • can it be done fast in c++ before it is loaded?

[cvvergara]

I will number one:

1. processing command line arguments mentioned here Program Options: -f --file <filename> #45 (that includes a lot of possible arguments allos mentioned in issues)
   It gives an idea of what is needed.

[cvvergara]

The "how to" for reading the data, I think is better with Osmium.
http://osmcode.org/libosmium/manual/libosmium-manual.html
which cover reading data from osm, pbf and opl formats.
A parameter option for the type of file is not needed as osmium detects wich kind of file it is dealing with.