Forum OpenACS Q&A: bug in openacs 3.5.2 ?

Posted by Eric Van Buggenhaut on
<debian maintainer's hat>

i'm currently trying to solve a critical bug someone reported months
against the Debian package of openacs. It'd be nice to solve it, since
tagged 'critical' and the package won't make to Debian/stable 'till I
get this

I can't join bug reporter by email (doesn't resolve) and don't know
what he
really means :

"Default install without presence of geotables causes chown -R var/ I
testedthis 4 times."

I don't know what this guy mean by 'geotables'. Can anyone help ?


</debian maitainer's hat>

Posted by Jonathan Marsden on

Geotables are the first part of the OpenACS 3.x datamodel to be installed.  See the script www/install/load-geo-tables and the related *.ctl files in that directory for the gory details.

I'm guessing that whoever posted the bug report probably meant that if the .deb's were installed in a certain way ("without presence of geotables"), then the install script(s) accidentally chown /var rather than a subdirectory of /var.  That would quite reasonably be considered a critical bug, since it could cause either major security issues or some other subsystems to fail to work correctly.

What I don't know is how it would be possible to install OpenACS 3.x without installing geotables.  Are they two separate .debs in the Debian installation approach?  If so, can the OpenACS .deb be installed before the geotables one?

Or, does the .deb install script have a faulty test for whether geotables are already installed, and go on to do bad things with /var if it believes geotables are present when in fact they are missing?

Jonathan (your fellow OpenACS packager!)