QA:Summer Coding Ideas - AutoQA detect false positives
AutoQA - add support for detecting false positives
|IDEA||nobody||nobody||python, searching best solutions, working with big data-sets|
|Generally this task would be about going through all the false positives AutoQA reported and figuring a method how not to run into the issue again and how to detect, if something changes. AutoQA needs to filter out this issues and not report them.|
Take a look at these issues (for example):
- add some black/white-list
- "explicit-lib-dependency liberation-mono-fonts" thinks that liberation-mono-fonts is a library, but it is IMO OK to depend on the specific font
- see the "libgnome" issue - somebody could add names of all packages and their's files into the dictionary
- problem "doc-file-dependency /usr/share/doc/gnome-python2-gnomevfs-2.28.1/async-xfer.py /usr/bin/env" could be reported directly as a bug. Could some Python-guy confirm that "/usr/bin/env python" was obsoleted by "/usr/bin/python"?
- migth ignore conflicts like this - like if I have conflicting basepackage-subpackage1-ver1 and basepackage-subpackage2-ver2 - I can determine, that the conflict is only theoretical, because if I have updated system, there will be only one of the packages.
- I do not know exactly what is the purpose of the test as the conflicts are valid properties of the rpm. On the other hand, file conflicts are bad. But take a look at the items found. It looks like there are just files moved from one sub-package to the another, like this (check different versions): `libmlx4-devel-1.0-4.fc12.x86_64 libmlx4-static-1.0.1-5.fc12.x86_64 /usr/lib64/libmlx4.a` or sub-package were renamed: `openscada-Special-FLibSYS-0.6.4.1-9.fc12.x86_64 openscada-Special-FlibSys-0.6.4-4.fc12.x86_64 /usr/lib64/openscada/spec_FLibSYS.so`
- ...and more.