[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: brltty: does this really need to depend on libicu?



Frans Pop, le Thu 23 Oct 2008 14:22:06 +0200, a écrit :
> On Thursday 23 October 2008, Samuel Thibault wrote:
> > > we could write some code to parse UnicodeData.txt (which is about
> > > 1MB).
> >
> > Which is basically what libicu is meant for <grin>
> 
> But why does libicu have to be so insanely big if its function is so 
> simple?

Because there is a lot of information in UnicodeData.txt, and
ICU includes it in libicudata in an time-efficient way (not
space-efficient).

> And if brltty only needs a fraction of its functionality, then I 
> think it should be reconsidered whether libicu really isn't the right 
> solution for it.

ICU is definitely the way recommanded by unicode.org.

I've checked glib, libunicode, libgucharmap, and libuninameslist,
only libgucharmap (3.5MB, only used by gucharmap gnome-applets and
abiword-gnome) and libuninameslist0 (3.3MB, only used by fontforge)
provide unicode character names. The line breaking information could be
provided by glib or libunicode. On the other hand, ICU is used by a lot
of applications.

So yes, we can use libgucharmap or libuninameslist to get the character
names, but I'm afraid (and actually hope...) that sooner or later other
packages from the small images will use ICU to properly handle unicode
too.

Samuel


Reply to: