#13184: Infinite loop with bash/readline/ICU on non-BMP Unicode characters
-------------------------------+----------------------------
Reporter: jessicah | Owner: pulkomandy
Type: bug | Status: new
Priority: normal | Milestone: Unscheduled
Component: Kits/Locale Kit | Version: R1/Development
Resolution: | Keywords:
Blocked By: | Blocking:
Has a Patch: 0 | Platform: All
-------------------------------+----------------------------
Comment (by pulkomandy):
From what I remember (we can meet on IRC later if you need me to dig more
into things):
- Our locale implementation relies on ICU to perform most of the work
- There are some tricks so that libroot can do this without linking
directly to libicu
- There are also some hacks to get gcc2 to have some wchar support, and
some rewrite of our glibc involved at the time to get this going. This is
our own code and there can be bugs in it (unfortunately widechar C API
does not seem to get much use)
Things to look for:
- How contexts for ICU are allocated/cleared when using the C APIs
- Making sure that ICU does not try to use C library functionality to do
its things, resulting in infinite recursion or other weirdness (this
happened at least once)
- Making sure we didn't break anything when backporting ICU to gcc2 (there
is a quite large patch at haikuporter for that)
- Making sure ICU itself works when used directly. It needs the relevant
data files, locale settings, etc.
--
Ticket URL: <https://dev.haiku-os.org/ticket/13184#comment:4>
Haiku <https://dev.haiku-os.org>
Haiku - the operating system.