Graham Seaman wrote: > I'm tinkering with new schema as well as including the standard > eduPerson Perhaps best to get it working with the standard before changing anything? > and was hoping to avoid having to strip out all the data and > then repopulate each time I make a minor change to the schema. That should not be neccessary unless your new or altered schema removed or redefines attributes or classes that are already in place in the directory. I added three or four new schemata to the directory I just installed, including EduPerson, and each time the data remained in place but I became able to add new attribute types to existing directory objects What order are you loading the schema files in? (Controlled by the two digits at the start of the file name) > I'm > populating it from a large Active Directory by script, which already has > quite a long run time. Pretty much exactly what I'm doing! > But I don't have any users in the directory at all yet, which is why I > was a bit surprised at the behaviour. I thought at least adding new > users with a new schema wouldn't be a problem. I guess if that is out > the next thing I need to check is what happens if I add a new 'may' > field to an existing schema - will it force me to drop all the old data > to install that, too. I do not think it should not do this at all. As far as I know adding EduPerson (or any other new schema) ought not to change what is in the directory already as long as you do not delete or redefine old classes or attributes that are used by existing entries. Are there no error messages at startup? Does the "new schema" you say you are "tinkering with" contain any attributes or classes with the same names or OIDs as ones in any other schema? Does the version of the eduPerson schema you are using contain a "changetype:" or any "add:" or "delete:" attributes? (I had to strip them all out to get mine working because Fedora didn't like an attempt to modify things that didn't exist) I wouldn't want to bet on what happens if a class is defined in one schema, then referred to in another, then redefined in a third!
