Improve performance of DIFF files

[eggwhites]

Thomas made some code changes here - Robert to integrate. Notes:

• Disable the feature class extension
• Using an update cursor as opposed to the search and store mechanism

[ThomasEmge]

So far the extension is still enabled. Changes would need to be made to the download, extract, symbolize toolset.

[tonyemery]

Is it normal that this tool is very slow to update the data? For example, I have a data set of 14,000 nodes I update often enough. If I use this tool, it takes several days for the update ... Or is it the fact that I use it in the wrong way and may be the tutorial of this tool must be enhance with screenshots. Do you have feedback on the people who use regularly and with satisfaction this tool?

[eggwhites]

Hello Tony - will you provide some further details on the area you are keeping up to date with the Load DIFF files tool (i.e., the bounding box, or area name if you are doing the initial lode from something like geofabrik), and the time interval at which you are applying the updates (i.e., every day, two days, week, etc.)?

[tonyemery]

In fact, I use this tool to upgrade datas loaded from XAPI tool. The extent is 44.19 N, 43.98 S, 4.70 W, 4.96 E and I apply apdate twice a week. My tool's options are : OpenStreetMap planet diff files base download URL : http://planet.openstreetmap.org/replication Input OSM Feature Dataset : V:_GDB\OSM_CCPRO.gdb\barrier Start Time for Diff Files : [nothing] Load updates only inside current AOI : [yes] Verbose geoprocessing logging : [no]

[eggwhites]

Thanks Tony - I did the DIFF files workflow this past week. I loaded the extent you indicated above, and after loading, made two copies. The first copy, I waited one day and then ran the Load DIFF files with no Start Time indicated. It did take a long time to complete - a little under 8 hours. When I ran the test against the second copy of the data and set the Start Time to just after the timestamp for the XAPI download finished (i.e., 2015-05-21T00:02:51) the DIFF files tool completed more quickly, in just under 2 minutes. I think you are right about the need to improve the Help documentation; the help indicates that the last hour will be used for the DIFF, but in reality I believe there is a query process that is clearly much more time consuming. Will you try using an ISO 8601 formatted time and let me know if that works better for you too? I'm also working on doing a compare between the resultant feature classes to make sure that the differences applied in the second test are at least as robust as the difference caught by the first.