I've grabbed that and converted it. It might have been better to start with an epub rather than a PRC. But, looking at the generated code, there are no problems with it. It does use "div"s instead of "p"'s for the paragraphs, but I don't think that is the problem.
I went straight to that line and selected it. I didn't get the same missing underline. It was just the last couple of characters. And when I tried to save it, the highlight didn't save. This was the first selection I tried.
So, I did some experimenting. Are all the problems you are having when you select the end of a paragraph? If I make sure the selecting doesn't end at the end of a paragraph, then there is no problem. If it does end with the end of a paragraph, sometimes there is. And it seems to happen more often if the paragraph ends with two punctuation characters.
Unfortunately, it doesn't seem to be as simple as "select to the end of the paragraph". It seems that if I stop the selection at the right place, the highlight will be saved. I don't have a feel for where this spot is.
I also tested with a purchased kepub. I was able to reproduce it there as well. This happens to be DRM free, so I was able to look at the code. Again clean and this one has "p" for the paragraphs.
So, I'm down to there being a bug in the selection engine of the kepub renderer. There is a problem with the spaces between sentences and hence the spans when the extended driver creates the kepub. That can be fixed. But, the rest is up to Kobo to sort out.
|