A couple of weeks ago I reported on my experiences loading users with the mod-users endpoint. Tod Olson suggested that I try the same size files with mod-user-import to see if there were any differences in timing, etc.

I haven't been successful in doing any such testing for reasons I will enumerate.
With users in particular, I like to do "in-and-out" testing -- put 'em in, take 'em back out. When I create the json for the user records I assign a UUID to each user and make a list of those ids in a separate file so that I can delete the whole batch after I've put them in, particularly if there's a general error that needs correction. Well, that's very difficult to do when using mod-user-import, because that module doesn't "respect" the uuid that is submitted as part of the json. It assigns its own id to identify the user, even if there is an id submitted in the json. So, submitting a batch of 5,000 users, say, and then running queries to get back the userids that were assigned so that they could be deleted would be a real pain, so I haven't gone ahead with a large load.

One of our developers here, William Welling, is going to raise the issue with the product lead to see if it can be addressed so that if a uuid is submitted with the user json it gets assigned to the user record.

Meanwhile, Chris (Manly), would you please remind me what the plans were for creating a module to do batch updates of existing users as opposed to batch adds? Was it going to be part of mod-user-import or another module? 'Cause mod-user-import has no update (put) method, just a post.

Anne



Anne L. Highsmith
Clinical Associate Professor
TAMU Libraries
5000 TAMU
College Station, TX   77843-5000
979 862 4234