Re: 3.5 PATCH: Fix Tru64 UNIX V4.0F libjava bootstrap failure innatFileChannelPosix.cc

[Per Bothner <per at bothner dot com>]

Michael Koch wrote:
> > Rainer>    MappedByteBufferImpl *buf
> > Rainer>      = new MappedByteBufferImpl ((RawData *) ((char *) ptr +
> > align), Rainer>  				size, mmode == 'r');
> > Rainer> -  if (ptr == MAP_FAILED)
> > Rainer> +  if (ptr == (void *) MAP_FAILED)
> > Rainer>      throw new IOException (JvNewStringLatin1 (strerror
> > (errno))); Rainer>    buf->implPtr = reinterpret_cast<RawData*>
> > (ptr);
> > Rainer>    buf->implLen = size+align;
> >
> > ... why are we checking `ptr' after creating the new
> > MappedByteBufferImpl? It seems more efficient and correct to check
> > first.
>
> That is Per's code. Her is probably a better person to ask.

No reason that I can think of.  While I don't think it matters,
it certainly does seem cleaner to check ptr first.  So feel free
to change it.
--
	--Per Bothner
per@bothner.com   http://per.bothner.com/