This has two additional benefits: 1. the beginner will stop be a beginner at some point, will be informed about the target format they are trying to use and will know how to save, save as, etc so this issue will stop be a problem for them and 2. it will get rid of the useless and annoying "do you want to save" warning whenever you want to create a non-XCF image file (because you already saved with "Export", you just didn't save the image as an XCF).
The current approach makes the (wrong) assumption that everyone works with XCF files and only exports to other formats. This might be true for some cases, but a lot of people want to work with other formats directly.
GIMP since version 2.8: only allow exporting to JPG/PNG/TIFF etc., warn if data wasn't saved. Actual results: only a few cases of lost data.
We did the math. Now you do yours :)
Here is something that actually happens since version 2.8: if you do not treat XCF files as the master image file format, GIMP complains all the time.
I'd like to see you prove me wrong.
That's way more annoying than the current GIMP implementation. I sometimes export the same image a lot of times (to work on reducing gif file size etc.) It's very convenient that I can have a master XCF copy and one-click export without any additional dialogs.
Also JPG is a lossy format, so it doesn't support even saving a one-layer file correctly. Do you really want a supposed "photo editor" to warn people every time they export to JPG?
I'm not even sure what this even means. No matter what file you're editing, you're converting it to a gimp internal format when you open it, you're working with it in a gimp internal format, then if you "save" it back to the format the image started in, what you're saving is a conversion from the gimp internal format.
The UI makes that clear. What people are arguing for is to make the UI less clear, not to change anything technical.
Replace PNG with any other file format (outside of XCF, of course).
The technical side is irrelevant, i am talking about the UI and its behavior here, not how it could be implemented.
I can imagine that they made the change due to many beginners being confused about having lost editing capabilities by storing their work as a png. The developers probably spend quite some time on the fora where users report these kinds of questions, so I don't think it is all that arrogant of them to believe that they know what causes confusion and what doesn't. Thinking that as a single user you are more knowledgeable on what is better for the over all ux than the actual developers, that I do think could be seen a little bit arrogant.