aspell incorrectly filter out text in --mode=email : it skip lines with "cyrillic small letter o" found in first 10 line characters. This happens because aspell incorrectly match this "small o" with ">" symbol, which is correct "email quotation mark" (in --mode=email aspell skip lines with ">" symbol within first 10 characters of line). As far as I understand, these two symbols match because aspell incorrectly work with Unicode (it cast <int> to <unsigned char>), look at symbol codes: 003E GREATER-THAN SIGN 043E CYRILLIC SMALL LETTER O I've created patch. It's worked, but I'm not sure is it doing right thing because I don't know C++ at all. :^) Reproducible: Always Steps to Reproduce: 1. 2. 3.
Created attachment 29946 [details, diff] patch... :) Add to .ebuild something like this: src_unpack() { unpack ${A} cd ${S} epatch ${FILESDIR}/${P}-quotechar.patch }
Does this still apply to aspell 0.50.5-r2?
Yeah. :-( This bug exists in 0.50.5-r2 too. Looks like my patch can be used for 0.50.5-r2 without changes and solve this issue again. I've reported this bug to aspell developers long time ago, but looks like they don't really do something with this issue. You can track my report here: https://sourceforge.net/tracker/?func=detail&atid=300245&aid=941238&group_id=245
Very good detective work. I've checked your patch and the source code and it does indeed look correct. Although it looks like from the source that they didn't plan for UTF-8 support at all :-) Anyway the patch has been included in aspell-0.50
