On Mon, May 09, 2011 at 08:02:06PM -0400, James Vega wrote: > On Sun, May 08, 2011 at 09:50:42AM +0200, Christoph Biedl wrote: > > Dominic Hargreaves wrote... > > > > > I can't promise to do anything with it in the near future, but happy > > > to grab a copy of it from somewhere if you can easily make it > > > available, to work on at a later date. Feel free to drop me a line > > > in private if you prefer. > > > > For the time being I've at least archived that copy. > > > > FWIW, amd64 is appearently not affected: > > That makes sense given that, I think, this is related to the > -Duse64bitint change in the Perl 5.12 packaging. On 32-bit systems, the > Perl 5.10 packaging wrote out Storable data in such a way that the new > use64bitint Perl 5.12 Storable can't read back in.
Just tested with Perl 5.12 built without -Duse64bitint and the file Storable creates indeed can't be read by 5.12 with -Duse64bitint. This is also documented in the "Storable Magic" section of Storable's POD: $info = Storable::file_magic($filename) … intsize, longsize, ptrsize, nvsize These are only present when netorder is FALSE. These are the size of the various C datatypes of the Perl that created this image. These must match the current Perl for the image to be readable by Storable. The odd bit is that git-svn seems to have been using netorder (via Memoize::Storable's nstore option) since the use of Memoize::Storable was introduced (git v1.7.0). -- James GPG Key: 1024D/61326D40 2003-09-02 James Vega <james...@debian.org>
signature.asc
Description: Digital signature