Good Morning Everyone, Not sure if this is the correct place to post this but it doesn't seem like a JOSM problem or a Newbies issue either. Maybe someone can direct me elsewhere if this isn't the best place.
OK... so I'm slowly plodding along importing CanVec data into south central BC. Yesterday, however, I encountered a problem that may be a result of my workflow process and I need some guidance in order to help avoid the same thing in the future. For the past 5 or 6 days, I have been working on 82L3.1.2 - a file that covers the southern part of Vernon, BC as well as parts of Okanagan and Kalamalka Lakes. There was already some existing OSM... some of which I had Bing-mapped myself several months ago and bits and pieces by several other account holders. As I could see this was going to be a particularly painstaking import, I decided to save the OSM download layer locally on my machine and merge item by item from the CanVec layer down onto it. By yesterday, the new composite OSM layer was looking pretty good. Considerable data had been added and some old OSM... such as my previous Bing mapping, had been deleted and replaced with the CanVec. Also, I had made other edits such as removing abbreviations, added turning circles, etc. I saved it one more time locally, then uploaded it to the server. In the course of uploading, there was a synchronization error on one node... a gas station in the original OSM that I had repositioned. The error stated that I had version 1 of the node, whereas there was an existing version 2. I synchronized that data point only and from what I could tell, the upload proceeded normally. However, checking the server later, it was clear that the upload did not go well. Over 6,000 data points had been uploaded with no corresponding ways. Some pre-existing OSM... such as my earlier mapping that I had deleted from the working copy had not deleted from the server, resulting in duplicate ways. There really seemed to be no pattern to the failure. Some water features uploaded, some didn't. Some highways did, some didn't. There was also no time-related pattern. Errors were not related to just my earlier or later edits. It was an equal-opportunity failure across the data set. Some time ago, I experienced occasional upload errors where nodes were being uploaded but not the ways. I was advised on the list at that time to cut the batch size from 2000 points down to 500. I did so and until now have had no further failures. I was also advised to just try to upload again as sometimes that's enough. So I tried re-uploading last night but it didn't recover. I can't remember the issue exactly in that case, but by then what was on the server was in such a mess I wasn't surprised. In any event... by 2 AM last night the south Vernon area was in pretty good shape again. There might be a couple of minor road-network issues to fix and some land-use and stream features that need restored. I still have probably another day or so of cleanup to get it to where it should be. So where did I screw up? Clearly, downloading an area from the server and working on it locally for a week before uploading is not best practices and can lead to synchronization errors... although in this case, it's hard to see how the error associated with a single gas station node could lead to such widespread failure. But when faced with a synchronization error like this, is it best to synchronize the whole data set or only the offending points? Is that where I went wrong? And just what are best practices for importing where it is clear that the job will take a long period of time? Make a couple of edits and upload? Make a couple more edits and upload? Is this the best approach? Or is there something in my procedure or in the tools that I'm missing that will help avoid this in the future. Thanks in advance, Sam L. Kamloops
_______________________________________________ Talk-ca mailing list Talk-ca@openstreetmap.org http://lists.openstreetmap.org/listinfo/talk-ca