Closed marco-m-pix4d closed 12 months ago
@Pix4D/integration: ready for re-review. Thanks for @aliculPix4D for spotting the suble new bug of the subtle existing bug. Please follow the new commits with their comments, they explain what is happening and step by step also simplify the logic.
Don't forget to close:
124
125
To double-check nothing is missing:
since #124 is mentioned in this PR (or in a commit comment), it will be closed automatically:
while #125 is unrelated.
In my understanding, there is nothing to do. Am I missing something?
In my understanding, there is nothing to do. Am I missing something?
no, it was my mistake: I was thinking of #/123 and for the other #124, I didn't know, it will be closed automatically:
Closes #124 and PCI-3302.
Best reviewed commit-per-commit and looking at the commit comments, otherwise it will be impossible to understand the details, and we do need to understand them in this case. Each commit tells the story and carefully builds on the precedent, keeping code coverage 100% (this doesn't ensure anything but is better than not doing it).
Last commit is perhaps the most unexpected, were we "define the errror out of existence", in the spirit of the book "A Philosophy of Software Design" by John Ousterhout (Talk: https://youtu.be/bmSAYlu0NcY?t=1313).
I have a recollection that the reason for the existence of the error is a comment I made on a previous PR by Aleksandar, so I am undoing what I proposed, because I learned from experience.
I decided not to touch the jitter initialization on purpose, because I think that the code in this PR is more robust and still as simple as it can. I am open to discussion.