This includes optimisations to my memcpy (previously only in an unoptimised or handwritten assembler form, now slightly optimised by default with a bit of manual loop unrolling).
A taskbar program has been developed, this works okayish but doesn’t have the fastest interface with the server for detecting window updates, so it can be improved a lot still.
Stability issues persist, and it turns out it’s kind of difficult to debug things within a GUI because the freezing could start in any number of places. However I expect this is a simple issue like a deadlock choking the scheduler, just hard to tell until I refocus on debugging.
It’s worth pointing out that it’s very easy to do this kind of stuff on a stable kernel, so adding features can continue easily if I just avoid debugging. I may have to make a decision whether to demo my platform kernel-first or GUI/tools-first – perhaps it would be worth developing some of this as an add-on for other operating systems first and allowing the kernel to stabilise more slowly. However I’m still ultimately a kernel developer so the kernel will remain my long term focus.
Leave a Reply