OK, I've played around some more, and my impressions differ a bit from localj's (#2 in this thread):
The Good:
I don't know whether I have to thank lBook or Apollo XXI for this, but my V3 already came with the latest firmware installed. Nice.
The TXT parser is pretty damn good. I could just throw a PG TXT file with those line breaks at it, and it did the right thing. No reformatting necessary. This is different from localj's experience, perhaps due to a newer firmware version? After I formatted it as one paragraph per line with an empty line after each paragraph, it also just worked. It choked on a file with one paragraph per line but without an empty line between paragraphs, and displayed that as one continuous stream of text. But since I've never seen one of those out in the wild, this doesn't really detract from the praise.
The Bad:
The HTML parser left me a bit less enthused. Links to other files in the same directory don't work. Not entirely unexpected, but it would have been nice. That it doesn't know how to deal with 3-byte UTF8-coded punctuation characters is forgivable, that it doesn't know the "mdash" named entity (&mdash
and just prints it verbatim, OTOH, is not.
Seriously. Named entity. Hello? It's not like there is an endless number of those (there are, in fact, around 250 of them), and mdash is certainly not the most exotic among them. This needs fixing.
After loading the offending HTML into Book Designer and converting it to FB2, it worked, although the dash was too short now. I'll probably have to play around with BD a bit more.
The Ugly:
ZIP file support doesn't work. It doesn't even show the ZIP files in the book shelf. RAR archives work just fine, though.
WTF?!