|Report created on Fri Feb 5 22:08:52 2016.||back to main menu|
For explanatory information and summary details, see the notes below.
|portname||maintainer||build error logs||Problem Reports||Responsible|
|ports shown: 7||ports with build errors: 1||ports with outstanding PRs: 0|
|ports with either build errors or PRs: 1|
Clicking on each column heading will cause the report to be redone sorted by that column. Clicking again will reverse the sort.
The portname column includes links to a more complete overview for that port.
The maintainer column includes links to a page showing the status of all ports for that maintainer.
The build error logs column represents the list of unique errors noticed in any build environment (if any). The errors are listed alphabetically. Each entry is a link to a particular errorlog. (In cases where the same error occurs in multiple build environments, the latest errorlog is used.)
The list of build errors that are detected, and a short description of each one, can be found here.
The PRs (if any) for the given port are listed numerically in the Problem Reports column. Thus, for each port, they should also be in order from earliest to latest.
Currently, no effort is made to correlate any individual build error with any individual PR. They are listed in adjacent columns only for your viewing convenience.
The underlying technology of this report relies on trying
to extract information from the existing PR database
entries. These entries are entered by human users using the
send-pr command. As such, the quality of the
entries varies greatly.
The fastest, and easiest, information is gleaned from a PR entry whose subject line contains the port category and port name, separated by a slash. However, if this algorithm only flagged those, it would miss nearly 50% of the ports PRs, not to mention all the 'framework' PRs.
So, as an extension, various heuristics are used to
guess what it is the user really intended. See the
prQueryUtils.py for the gory details.
What's important to understand is this: there is
no possible algorithm that will correctly identify all
the ambiguous PRs without getting a few false identifications
and still run in less than geological time. So, before you
are tempted to file a PR on this algorithm itself,
read the code to understand its design tradeoffs, and
then consider instead filing followup PRs to the ambiguous
PRs that would disambiguate them instead.
Thanks -- the author.