Count Daily Active Users

[pthiess]

Provide data about how many people are using Brackets by parsing the S3 logs that serve the update JSON file.

Data to parse:

• Within a 24 hour period, how many people have used Brackets?

Need to take into account:

• Scrub out the pings to the JSON file from brackets.io visits

Can we use a service:

• s3stat.com
• cloudlytics.com
• qloudstat.com

Questions

• Do we remove the processed logfiles after we extracted the usage data?
• Is it enough to use a simple UUID<->IP mapping for anonymization? Do we need to keep this mapping between reports for the next report(s)?

Things to consider

• there is no easy way to generate custom reports without inspecting all the files again, let's say "show me the usage data between May 1st and May 31th". This requires to look inside the log files to get the timestamp.

Reference

Brackets Registry log file processor https://github.com/adobe/brackets-registry/blob/master/downloadStats/logfileProcessor.js

[dangoor]

I think we should clarify exactly what the output is for this. IP addresses don't factor in, really. We just need to count how many pings there were that didn't have a referrer of brackets.io over a 24 hour period. I hope s3 gives us the referrer!

We should have a JSON file that has the data broken down by date. We should consider putting that in git or otherwise rotating it in case the script goes awry some time. With this, we could calculate average daily users over some period of time and would not need to go back to the original log files.

[ingorichter]

The referrer is part of the log files. So this shouldn't be an issue.
We can save the extracted data in a json file and put save it in the S3 bucket or a git repo.
I'm going to use an Adobe internal S3 Bucket to store the json file.