Hi, Chris Lamb wrote: > I fixed a couple of segfaults that evening, but I hit one that eluded me > and went to sleep. The code is extremely brittle - I wonder whether we > should not just go with a newer upstream.
> Alternatively - Thomas, did you look already whether the rdepends > actually use this package? I remember at least qmtest had some pointless > Build-Depends which I NMU'd away before, and--if the other packages are > not failing to build--it is unlikely that they are actually using > ExtensionClass given the level of fail there. IMO it is critical not to have epydoc segfault (should probably get a Conflicts: even when python-extclass is removed). python-extclass as is seems useless: Even when fixed, which software would sanely use this? None, because any reasonable software would expect a reasonable version of it (except that Zope - the extclass upstream - label extclass as obsolete since python 2.2 anyways). Fixing zorp would amount to converting the extension module's python class to "new style" classes. I don't know how difficult this exactly is, but it is an opportunity to introduce bugs (don't know why upstream did not do that yet). Having a firewall that depends on the current extclass does not look attractive, either, though. I'm not sure qmtest should need the extensionclasses (not listed on upstream page with prequisites[1]). To me it seems completely reasonable to throw out both reverse depends: qmtest has never been released and zorp is not exactly maintained or popular. Kind regards T. 1. http://www.codesourcery.com/qmtest/2.4/download.html -- Thomas Viehmann, http://thomas.viehmann.net/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
