Am Sat, 08 Mar 2014 22:07:09 +0000 schrieb "Sean Kelly" <[email protected]>:
> On Saturday, 8 March 2014 at 20:50:49 UTC, Andrei Alexandrescu > wrote: > > > > Pretty much everyone using ICU hates it. > > I think the biggest problem with ICU is documentation. It can > take a long time to figure out how to do something if you've > never done it before. Also, the C interface in ICU seems better > than the C++ interface. And I'll grant that a few things are > just far harder than they need to be. I wanted a transcoding > iterator and ICU almost has this but not quite, so I've got to > write my own. In fact, iterating across an arbitrary encoding in > general is at least not intuitive and perhaps not possible. I > kinda gave up on that. Um, and using UTF-16 as the standard > encoding, requiring many transcoding operations to require two > conversions. Okay, I guess there are a lot of problems with ICU, > but it handles nearly every requirement I have, which is in > itself quite a lot. You find the answer here: http://userguide.icu-project.org/icufaq#TOC-What-is-the-performance-difference-between-UTF-8-and-UTF-16- In addition it is infeasible to maintain code for direct conversions with all the encodings they support. The project doesn't aim at providing a specific transcoding but all of them equally. What can you do. For Java it is easier to accept since they use UTF-16 internally. -- Marco
