Things to do¶

These are just some open issues that I am keeping track of.

Features to Add¶

• Create a cron method that runs all the phases one by one, stores the results to a log, and produces output suitable for cron. (e.g., nothing on success, useful info on failure)
• Fix the post and postgraphs module so that the graphs post will be a reply to the last regular toot.

Bugs to Fix¶

• Get rid of lookback and refactor around the start time of the event and the hours margin we're going to look at. Lookback was an early concept that is not the right way to do that.
• We might want to limit the top_n to only folks who allow public read on their server. That is, don't publicise or republish pointers to servers that haven't made it clear that they want to be publicly found. It's surprising that some of the non-local toots (e.g., from timeloop.cafe) still make it into the tops.
• Reimplement debug correctly as an INI file option, so different phases can have different debug levels

Done¶

• Done Add some multiprocessing to fetch() so we can work on multiple servers at the same time.
• Done Change analyse to write out intermediate files with deduplicated toots, metadata, start/resume, etc. Change other phases to read those processed result files.
• Done Build a bit of pause/resume. Store progress. Pick up where we left off
• Done Create the init function
• Done Get some automatic building of documentation from source code
• Done Create a method that posts the graph as a visual attachment with interesting alt text.
• Done Create a method that posts the word cloud as a visual attachment with interesting alt text.
• Done The "most replies" counts self-replies. So someone who starts a root post and then does a long thread will have many replies, and that's not really what we want. We want "most replies by other people."
• Done Add HTML generation of output runs (i.e., make a web page that writes up the run, has the graph images, etc.)
• Done Add output directory distinct from data directory
• Done Add support for stop words in the config, so popular phrases can be filtered out of the word cloud.
• Done Implement event_labels options better
• Done Fix bug on number of servers: say how many are left and get the server name right
• Done Count non-responsive servers so they can be removed from the count.
• Done Save each server to its own file. Build a loader that loads them all
• Done Handle toots that come from private servers better. We get many copies. figure out what is the Best copy and get rid of the others.
• Done Build a word cloud SVG from the content of the toots we find. Remove any silly emoji, icons, HTML, and other non-word things before generating it