it took me 1 month to fix this compiler bug

one pesky little bug.
one month of development.


more about SSA form:

introductory stuff:
- https://en.wikipedia.org/wiki/Static_...
- https://blog.yossarian.net/2020/10/23...

converting to ssa form:
- ssa form: https://www.cs.cornell.edu/courses/cs...
- dominators: https://www.cs.cornell.edu/courses/cs...
- and: https://www.cs.rice.edu/~keith/EMBED/...

converting out of ssa form:
- register allocation: http://web.cs.ucla.edu/~palsberg/cour...
- and: https://link.springer.com/content/pdf...
- phi elimination: http://web.cs.ucla.edu/~palsberg/pape...
- and: https://www.cs.cmu.edu/afs/cs/academi...

implementations:
- mine: https://github.com/leddoo/kibi/tree/d...
- negate's tb: https://github.com/RealNeGate/Cuik/tr...


chapters:
00:00-00:20 - intro
00:20-01:29 - the bug
01:29-03:14 - how python does it
03:14-05:29 - why mine is broken
05:29-08:53 - finding a solution
08:53-10:37 - copy propagation
10:37-12:35 - ssa form
12:35-14:16 - overview of the new compiler