4 ms·
It was interesting to me how many of these were related to making async calls to adjust state resulting in race conditions. In many cases the argument was made
by Yeroc 6y ago
It was interesting to me how many of these were related to making async calls to adjust state resulting in race conditions. In many cases the argument was made that async calls were not justified. Many times we follow a particular call pattern in order to keep things consistent without fully considering whether it's necessary and/or safe.
Also I wasn't aware of the tool named Frida[0] that she kept referencing so I had look that up.
[0] - https://frida.re/ https://frida.re/
- danlugo92 6y agoSolution is this: Assign UUID to each async call, including retries. Or disallow retries. Sprinkle in some optimistic update if you want. Keep track of results, mirror lower level timeout (75 seconds for HTTP).