The import progress is, in fact, inconsistent at times. Sometimes it goes backwards. Surely that’s a bug! No, actually it’s not. It estimates the percentage done based on the best information it has at the time. When it import starts, we know very little. We just have a reference to where to get the document. We don’t know how large it is or what type of document it is. We don’t make an estimate until we have the size and type. Surely we should be able to get a decent estimate then? We can get some good guess, but that’s about it. Take the example of PDF. The PDF can be packaged in the number of ways that affect the time to import it. It may be nothing a series of images, in which case, we have to run through a optical character recognizer (which is time consuming).
