I'll be happy to test this out, although I probably can't get to it till the weekend. Do you have a branch somewhere with all these changes made?
--Andrew Whitworth On Thu, Feb 26, 2009 at 5:57 PM, Reini Urban <[email protected]> wrote: > Good news, not for 1.0, it's rather huge, but for later: > > I found the last bugs on 32-bit reading 64-bit pbc, and would > appreciate countertests on 64-bit machines. > It's a rather big patch, because I also const'ed pmc_freeze, added > ptr_alignment checks, more #ifdef'd tracings > and added random documentation. > > But the two essential problems where wrong ALIGN_16 which I replaced > with a simplier and more general approach, > by calculating the padding in bytes and setting the cursor absolute, > not relative. > All pbc cursor padding uses now the same macros and approach, in > pf_items.c and in packfile.c. > > And: fetch_iv was not set for thaw_pmc, it was forgotten in a new > switch/if codepath of mine. > > I think I understand that code now pretty well. > > An unrelated bugfix: Parrot_compile_string missed to set interp->initial_pf = > pf > > What is missing: Finish some floatval transformers and a test matrix for > these. > cvt_num16_num8 is ok now. > I've marked the tested status in the code, since our smoke tests don't > catch these cross-platform regressions. > > -- > Reini Urban > > _______________________________________________ > http://lists.parrot.org/mailman/listinfo/parrot-dev > > _______________________________________________ http://lists.parrot.org/mailman/listinfo/parrot-dev
