Here are a few comments on the problems you listed:As noted, I have had many problems with projects over the years. Projects created with older versions that wouldn't load, projects where I lost data because some absolute path had changed, projects with process icons that no longer worked. Lots of people use them and that's fine. I don't trust them and I don't trust my data to them. And frankly, I don't see that they provide any benefits to my workflow. So I don't use them.
In this case, I was not suggesting they not be used. My point was that loading an old project into a PI installation that is having problems is adding one too many variables, and I think it's a bad idea
- Projects that entirely won't load are corrupted projects; this may happen if you interrupt the copy (then you'll miss files in the project folder) or if the files are corrupted on the disk
- There is no absolute path involved in project files, projects can be moved freely, but avoid loading them from a NAS or Dropbox folder, just ensure they are accessible from an internal HD (this is also needed for loading images, for example)
- The dependency on missing processes is the only thing that may affect the loading of an image in a project because the image state (its history) depends on something that is missing; we already had a discussion about that and how to address this case, just planned to be implemented, but it's nothing that blocks you from saving/loading projects today