enrico.scholz@xxxxxxxxxxxxxxxxxxxxxxxxx (Enrico Scholz) writes:
> The question is, whether it would be worth to begin such an
> implementation
FWIW, I expressed my ideas about the underlying database in an ERD at
http://www.tu-chemnitz.de/~ensc/fedora.us-build/qadb/qadb.ps
The role of 'pkgsubmission', 'comment', 'review' and 'reviewer' is
the interesting part. Basic procedure is given below in a technical
description (which is not interesting for the enduser):
As soon as a 'pkgsubmission' becomes valid for a certain 'os_branch', it
will be given to the buildsystem and be built for this 'os_branch'. When
build fails, an automatic 'comment' will be generated with 'REJECT' as
the 'review'.
A 'valid pkgsubmission' for a certain 'os_branch' is a 'pkgsubmission':
* which is not obsoleted by another 'pkgsubmission', and
* does not have a valid REJECT 'review', and
* which has a ranksum above X (value must be defined; currently it is 2)
A 'valid review' is a 'review' without a valid veto. This review must be
related to a direct 'comment' for a 'pkgsubmission'. 'reviews' related
to 'comments' of 'comments' are not valid and should be rejected by the
userinterface.
A 'valid veto' is a 'comment' of type VETO which does not have a
VETO. This applies only once, you can not veto the veto of a veto. To
do this, create a new veto for the original comment.
The ranksum is the sum of 'rank' of all 'reviewers' which did a valid
APPROVAL for the pkgsubmission.
Enrico
