Scripts for triaging and prioritising Rust issues
Everything interesting is in src. You'll need to copy config.json.orig to
config.json and fill out all the values. Run using run.sh in src (you'll
need to setup by running init.sh in staging first). To run on a server,
first use run.sh, then copy everything in staging to the server, run
init.sh, then nohup nodejs triage.js &. I think node.js is the only
prerequisite.
You'll need to setup your GitHub repo to send a webhook on issue comments and issues (which covers new issues and [de-]labeling).
API
/hook receive GitHub web hook. We record priority changes made by changing
labels (any label starting with P- or I-nominated). We also allow changing
labels in a comment using, for example, triage: P-high. The comment is then
also recorded. Users allowed to do this are listed in config.json
/data dump currently collected data as JSON.
/mail_digest trigger the digest email and empty the current data state.
Source overview
Most of the work is done in triage.js. call.js has some basic helper
functions for making GitHub API calls. digest.js produces the digest email
text (as html).
The current state of the program is pretty much all in the data global. This
is regularly stored to disk (in data.json) so we can recover from a crash.
We record all priority changes, either by label changes or by using triage in
a comment (the latter is preferred since the comment can add context). Sending
an email sends all recorded changes to the email address listed in
emails.json. It also saves a copy in the digests folder. It then resets the
current state.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent