Fwd: Memory error

[Erik Bray]

Jason Winnebeck <gillius-ml at gillius.org> wrote:
> Maybe it does something silly like assume sizeof(int) = 4 bytes? I don't know,
> on modern compilers was the decision made that int should become 8 bytes (64
> bits wide)? I know that size of pointers would have to change, but if it is
> trying to do IO based on pointer sizes that would be really out there.
>
> Jason
>
> David Blaikie wrote:
> > my mistake  - ssphys has the problem...
> >
> > On Nov 30, 2007 4:15 PM, David Blaikie <dbb14 at columbia.edu
> > <mailto:dbb14 at columbia.edu>> wrote:
> >
> >     eek.  This is the same error I was going to bring up next - the
> >     error I got when migrating a small database!  Looks like I'm out of
> >     luck, then!
> >
> >     Anyone have any idea why this should be so?  Why should vss2svn have
> >     trouble on 64?
> >
> >     David

I actually ran into this problem a while ago and never remembered to
post my solution.
It's pretty much as you say--in SSTypes.h, unsigned longs are being
used as members in some structs, breaking things, due to the
assumption that they'll be 32-bits wide.  I just changed the typedef
for "ulong" at the top of the file to be an unsigned int, recompiled,
and it worked fine.

Erik