Something fishy with PersistentDimSpace's content management

[Antti-Juhani Kaijanaho <gaia@xxxxxx>]

Hi,

I've been tracking down the problem(s) with conversion as reported by
Marlene.  It seems that the current (CVS HEAD) versions of ZZPersistentDimSpace
& co have a strange bug regarding reading the persistentdimspace content data.
The cells are all there as they should be, but getText() and even ContentHash's
get() return null or "".  If I brute-force-enumerate all TreeRecords in the
content hash, I find that all the data is in there, but get just does not
return it.  However, CVS as of 2000-07-06 does parse the same space as if there
was no problems.

I've been diffing the code and have not seen any obvious bug in the code.

Any ideas?

Note that you need to create the test space with GZigZag from 2000-07-06
(apparently any such space will do).  Current GZigZag with -oldspace does not
create spaces that trigger this bug.

The -D switch with cvs checkout helps if you want to get the old code.

-- 
%%% Antti-Juhani Kaijanaho % gaia@xxxxxx % http://www.iki.fi/gaia/ %%%

                     Keep the Deja Archive Alive!
        http://www2.PetitionOnline.com/dejanews/petition.html