Log file Viewer

[02:58:38] <ztane> mitsuhiko: the point in the post is valid, but the problem is that it assumes python was developed and is used for mainly web apps. However, could it be possible to write an extension, utf8str that is a subclass of str/unicode but would use utf8 as storage only?

[03:52:41] <dash> ztane: why do you think it assumes that?

[06:51:40] <ztane> mitsuhiko: also, in cpython3.3 C-api now there is PyUnicode_AsUTF8String which caches the representation, so one could get half the advantage by just writing the join() in C

[07:02:59] <dash> heh

[07:48:48] <ztane> as far as I read correctly the source at: http://hg.python.org/cpython/file/52fbc7bb78ad/Objects/unicodeobject.c

[07:49:24] <ztane> the code is a bit inefficient for onetime encoding of utf8, but if you reencode the same string over and over again to utf8, it should use the cached version