The logs.tf parser is... buggy. Unfortunately, it doesn't seem like zoob is working on it any longer. By supporting uploading logs, we can improve the parser and have better results. This can also allow for new stats to be displayed.
This will probably start with BV's parser, although icewind's could also work. We should match the logs.tf API to ease integration. This means generating/storing API keys. For a prototype, we probably don't need a web interface for this. However, eventually this should be done through #17.
According to BV, his parser takes around 40 seconds to run. This is fast enough that we can parse logs in the foreground while responding to requests. The longest part will probably be doing the actual import. Maybe we can redirect to an interstitial page while we wait for the log to be imported? We will also need a new table to store the raw logs.
This will depend on #84, although it would also be possible to re-parse logs for additional information. This may be much more pragmatic.
The logs.tf parser is... buggy. Unfortunately, it doesn't seem like zoob is working on it any longer. By supporting uploading logs, we can improve the parser and have better results. This can also allow for new stats to be displayed.
This will probably start with BV's parser, although icewind's could also work. We should match the logs.tf API to ease integration. This means generating/storing API keys. For a prototype, we probably don't need a web interface for this. However, eventually this should be done through #17.
According to BV, his parser takes around 40 seconds to run. This is fast enough that we can parse logs in the foreground while responding to requests. The longest part will probably be doing the actual import. Maybe we can redirect to an interstitial page while we wait for the log to be imported? We will also need a new table to store the raw logs.
This will depend on #84, although it would also be possible to re-parse logs for additional information. This may be much more pragmatic.