There are two directons.
First, statistics (per kernel version):
kernel size distributions (histogram say)
options frequencies
Second, logs analysis: in the database, there are several "logs" messages (about build failures, warnings, etc.)
We would like to group together "closed" logs messages (and display them) per kernel version
For the two directions, the computation can be heavy and should be done "offline".
I propose we rely on a notebooks technology (https://jupyter.org/)... and the notebook will be refreshed every day and integrated within the website.
There are two directons. First, statistics (per kernel version):
Second, logs analysis: in the database, there are several "logs" messages (about build failures, warnings, etc.) We would like to group together "closed" logs messages (and display them) per kernel version
For the two directions, the computation can be heavy and should be done "offline".
I propose we rely on a notebooks technology (https://jupyter.org/)... and the notebook will be refreshed every day and integrated within the website.