🚀 go-pugleaf

Simon Waters <Simon@wretched.demon.co.uk> writes:

> Goswin Brederlow wrote:
> > I fixed some compiler warnings and an error in the Rand32() function
> > (must be int on 64 Bit cpus, not long) and wrote a converter for the
> > book.dat file to/from host byteorder/padding to a general packed big
> > endian byteorder. Save a lot of time during compilation for Debian,
> > because one book now fits all architectures.
> > 
> > Intrested?
> > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=129653&repeatmerged=yes
> 
> Always interested - I'm doing this to learn these things -
> mostly I've think I've come to appreciate Java better so far ;)
> 
> As of last week there aren't any compiler warnings in the CVS
> version at Savannah (finally). So probably some duplication of
> effort going on here.

I will take a look at the cvs when this compile run is done, which
means about 10 days or if I find time inbetween.

I found some stupid bugs in the conversion, which made it a nop. I
fixed that now, url stands as it is once it updates. I tested it with
a much smaller book and that now worked.

I'm doing a full compile on different archs now to compare the result
of the native compile, conversion to general form and back. Does
anyone have a middle_endian system? I'm ignoring them for now.


I also experimented with using a GHashTable (from libglib1.2) instead
of a simple array while building the book. Doesn't seem to make much
of a dent in the speed, but it should improve the bigger the book gets.

MfG
        Goswin