libiconv - final summary

Gabor Kovesdan gabor at
Mon Aug 17 14:35:53 UTC 2009


the last week I experimented a bit with the regression test I wrote and 
reduced the size of the generated data because this was the reason why I 
haven't committed the reference data to p4 yet. Also, I've changed to 
using library instead of statically linking everything. Th iconv stuff 
will be part of libc, so this layout seems more real. Some problems have 
come out which I'm still working on. Good news is that probably I can 
keep working on this project as a BSc thesis, which will allow me to 
dedicate more time and go into deep details, review performance, etc.

Some words to resume the progress. Almost all encodings, which are used 
in our locales are supported. I've also added some extras. There's a bug 
in Big5 module, which makes it segfault. There are two (iirc) missing 
encodings from the CP family. The foward encoding (to UTF-32) is of the 
same quality as GNU iconv and GNU libc-icov but the reverse encoding 
sometimes less punctual, there isn't such a good level of 
transliteration. I'll need to look into it in detail.


More information about the soc-status mailing list