On Sat, 20 Oct 2012 15:38:35 +0100, John-Mark Bell wrote: > I'd rather we didn't do that, as it somewhat defeats the point of having > a controlled build environment.
Fair enough. > Either we need to work out what causes > the newlib iconv implementation not to work in our environment, I spent a lot of time on this in the past, trying to construct a test case which caused newlib's iconv implementation to crash. I couldn't do it, even feeding data that caused a crash via NetSurf/lpu wouldn't crash newlib when called directly. > or get libiconv building for the ppc-amigaos SDK. That's probably easier. I last built it a few years ago, IIRC after building the library it starts building other things that fail, but the library itself builds and works fine. Chris
