Originally Posted by DNSB
One of my co-workers has had issues with some corrupted epubs on his Sony reader. Unlike the Kobo, it did not lock up but rather spread the corruption to files loaded after the corrupted file so you got multiple books that were garbage when you attempted to read them. He also ran into an issue a few years back where some epubs converted by Calibre caused his reader to reboot though that was fixed a couple of Calibre releases later.
Personally, I find that using epubcheck or the version of Flightcrew's epub validator built into Sigil will detect the majority of corrupted epubs before you sideload them.
Sigil will make changes to the structure of the epub file when you save the file but those changes make the epub conform to the standard epub structure -- the mimetype file alone in the root of the zipped file, the META-INF folder with the container.xml file and an OEBPS folder with it's two files for toc and content and subfolders for text, styles, images, fonts, etc. so the structure change does not bother me.
If it's that well formalised, and checkable, sounds like something Calibre desperately needs to conform to, and implement - of course, it may do this already, I wouldn't know, as everything from Calibre has been accepted by my KT up to now.