Gas regressions: reverse outputs
Context
When reports of a full benchmarks run are sent, the most recent values are found on the right of each line, which means that we have to scroll all the way to find the interesting information.
This request proposes to reorder the columns so that the most recent values are found first (i.e. on the left).
Manually testing the MR
Let's download the raw results of some benchmark runs: input_csv_dir.tgz
.
Also, not ideal, but let's download a modified version of watch_regressions.sh
.
Extract the contents of input_csv_dir.tgz
and add execution rights to watch_regressions.sh
.
Then, let's run watch_regressions.sh
before/after the request. First parameter is our Tezos local repository, second is the input CSVs directory (full path). For instance, if we downloaded the files to /tmp
:
cd /tmp
tar xvzf input_csv_dir.tgz
chmod +x watch_regressions.sh
cd ~/Work/tezos
git checkout master
/tmp/watch_regressions.sh ~/Work/tezos /tmp/input_csv_dir
git checkout naih@gas-regressions-rev-outputs
/tmp/watch_regressions.sh ~/Work/tezos /tmp/input_csv_dir
The script tells us where the output files are. We can check that the alerts
files are the same, and how selection.csv
have their columns inverted compared to each other.
Checklist
-
Select suitable reviewers using the Reviewers
field below. -
Select as Assignee
the next person who should take action on that MR