Another day another fight with Maestro.

Today's battle started when I noticed that one specific test was flaky to the point of making development unbearable.

After a full 8+ hour session I've reduced it down to Maestro unreliably traversing iframes in WebViews: github.com/mobile-dev-i...

My gauntlet testing is still failing after 2-3 runs, but as I can now reliably detect a failure due to iframe traversal I'm able to simply ignore them in the stats. Will leave it running for a while to see if there is any remaining flakiness. Then continue dev.

Sidenote: An interesting part of the day was adding randomly lengthed cryptographic padding to my RSS mocks as my EPUB generator was turning out .zip's at exactly the same file length... which was hitting my rather too crude cross device archival synchronization logic.

Some day I'll move it to using digests rather than just file sizes, but not today. Realistic use cases won't hit this corner cases anyway.

runs off to find some wood.