[Metakit] PyDS 0.7.2 database corruption

[Georg Bauer]

Hi!

Am Freitag, 14.05.04 um 21:26 Uhr schrieb Jean-Claude Wippler:

> I suspect that space characters in the description string could be the 
> problem.  They will restructure the datafile, causing data loss (but 
> not really damage - the loss is because properties will be dropped).  
> As for the 100% CPU - the only case where I have seen this before is 
> with (sub)views containing no properties.

PyDS doesn't have any spaces in getas description strings. The sample 
was from a different context and a different project (PyCS - and even 
that has no spaces, the spaces where introduced by Yasushi Iwata when 
trying stuff).

So Nicholas problems are not due to spaces. And my broken PyCS database 
isn't broken due to spaces, too. The problem with the PyCS database is 
that it's too large and to much private content as that I could upload 
it somewhere - the file includes all users settings, passwords, private 
data etc.

The symptoms are allways the same: the program grabs all CPU resources 
and the process itself hangs on opening a view. Opening the database 
file itself works fine. Just the view opening hangs. Both projects 
(PyDS and PyCS) are written in Python. They still usually use Metakit 
2.4.9.2, as Yasushi had bigger problems with crashes when trying to use 
Metakit 2.4.9.3 (and I myself use Debian packages that still are at 
2.4.9.2).

The mkstats didn't work for me as reported in one of my earlier 
messages, there were some problems with missing symbols in glibc - 
don't actually know why, as mkstats is a statically linked executable 
and shouldn't have dependencies on shared libraries on my system ...

bye, Georg