Perhaps the TL;DR for me is that the gains in time don’t seem particularly large. Based on those figures I project the effect on overall bootstrap time would be quite small. Do we have any figure on that?
If that is correct, then I think we should go for the simplest solution possible, which I think is:
- Re-enable landing pads on stage1, so that things work in a uniform fashion.
- Refactor things so that we can disable landing pads everywhere
- this presumably means using the custom panic handler stuff for rustc to print a nice msg on an ICE?
- I’m not really sure what it means to refactor the test running infrastructure here