[postgis-devel] different EWKB representations on different machines?

[Alex Mayrhofer]

strk at refractions.net wrote:

> 1.0.0RC1 is known to contain memory alignment problems.
> If you can reproduce bugs with 1.0.4 please come back on the issue.

upgraded to 1.0.4, still the same. PostgreSQL was compiled with 
--enable-cassert, but does not report any problems at all:

query:

testdb=# select pointn(line, 2) from test;

("line" is a 4D geometry with about 800 points)

gdb output:

Program received signal SIGBUS, Bus error.
0x7107a660 in ?? ()
(gdb) bt
#0  0x7107a660 in ?? ()
#1  0x7107a610 in ?? ()
Previous frame identical to this frame (corrupt stack?)

(postgres, as said before, does not report anything)

looks to me like something while traversing the geometry goes horribly wrong 
here. how can i help tracking this down? start printf debugging?

cheers

alex