Angus Leeming <[EMAIL PROTECTED]> writes:
| Lars Gullik Bjønnes <[EMAIL PROTECTED]> writes:
| > Note that it will most likely only work properly on linux.
| > (I use wchar_t in a couple of places and that changes size on
| > windows, also I use iconv quite heavily for the moment.)
|
| I've been watching your progress on cvslog; I like the way it's looking. All
| those ucs2_to_ucs4, utf8_to_ucs4 methods are extremely intuitive. However, I
| note that:
| 1 They don't conform to your naming convention. I find this especially funny
| after the the hard time you gave Abdel about this ;-)
:-) It is not my naming convention. And I am not sure that I agree
with what is in the docs. I am perhaps to good at discussing with
different hats on. (If we begin the discussion with: "What naming
convention do we want?", instead of "What naming convention does our
rules say we should use?" then I might seem like a completely
different person.)
| 2 The code itself in many of these functions reinvents the wheel. It
| would be nice if ucs2_to_ucs4 et al were just iconv wrappers.
Aren't they just iconv wrappers?
In what way are they reinventing the wheel?
| Which brings me to your second point: what's wrong with using iconv quite
| heavily?
Nothing really, but it is probably a lot slower than just doing:
uint16_t ucs4_to_ucs2(uint32_t from)
{
// Add a check for out-of-bounds
return uint16_t(from);
}
--
Lgb
