- description modified (diff)
If lurch attempts to load a file, and for any reason the file cannot be loaded, attempt to recover sensibly. For example, if the file can be loaded into a Lob, but something goes wrong with the meaning of that, such as duplicate IDs, then still load as much as you can. In the case of duplicate IDs, just delete the duplicates. In the case of some other semantic problem, at the very least we can load the document with no bubbles at all. If even the Lob's XML is invalid, we can probably still load anything that is inside an OMSTR (and that's not within an attribute).
Be sure to make the recovered file have a different filename, so the user doesn't save back over their original. That way they can submit the original for us to look into for bug fixing, or possibly even recovering all their data.