You don’t do it that way. You figure out the how it’s gong to fail, when that failure is likely, and then engineer it not to do that in the relevant timeframe.
The original engineer was right and you are not.
The original engineer was right and you are not.
No comments yet.