Hi Roman, Roman Kennke <[EMAIL PROTECTED]> writes: > Hmm, I'm not talking about fixing the spec (I've read that bug report > while searching for clarfication on the spec actually). When the spec > doesn't tell _that_ the returned array is zero terminated, I think we > should assume that it isn't (and it seems to be the trend that the spec > should be clarfied by saying that an implementation isn't required to > return a zero-terminated array, but this is only speculation). What I'm > asking is, should we fix the java.io C code to deal with > non-zero-terminated jchar arrays? Unfortunately, this probably means > allocating additional buffers, because we really need zero terminated > strings here (AFAICS).
If you rewrite WITH_UNICODE_STRING to not use the runtime to allocate and deallocate the jchar array through GetStringChars and ReleaseStringChars but allocate and deallocate the jchar arrray yourself using GetStringLength (+ 1) and then fill it through GetStringRegion() it looks like you don't really need to allocate any additional buffers. Cheers, Mark
