What this is

The Problem: We have a problem on our project (http://ooici.net/): We're writing multiprocessing messaging-based code with a single logfile/stream, using Python logging. The problem is trying to decipher what's going wrong from the intermingled log messages.

The Idea: What you want to see is a message sequence chart, where there's a column in a browser for each process/thread, and the vertical axis is time.

The Technology: 1) html5 canvas, used as simple (x,y, text) 2) jquery (for event hooks, viewport size) 3) python + twisted to create a RESTful server 4) javascript to pull and display

The components: 1) ionlog.py to parse and restructure the logfile into a set of data with 3 methods a) get_config returns a dictionary of num_cols and column_names b) get_names returns just the column names c) get_single_log(name) returns just an array of messages for a given name Result is an array of dictionaries with i) delta_t time since first log message (relative time, in seconds) ii) timestamp, ISO 8601 (absolute time) iii) log level for colorization and filtering. DEBUG, INFO, WARNING, ERROR, CRITICAL iv) Msg has the string, filtered a bit for json-encodabilty 2) ws.py to present via REST and also handle JSON encoding 3) one.js for javascript code. a) Pull the configuration to draw the canvas b) add column headers c) for each column, draw the messages on the canvas 4) one.html as the page to load that invokes the jscript

Lots of work left!