|
From: Philipp K. K. <pk...@sp...> - 2025-01-12 15:52:06
|
I have two more fixes which I want to pick from the next branch to trunk before RC2: * https://sourceforge.net/p/sdcc/code/15221/ This should have been part of a previously picked fix, but I forgot to commit it to next in time. It is just a fix for a mistake in a regression test (but we need it to get rid of red dots on the snapshots page). * https://sourceforge.net/p/sdcc/code/15222/ This is the fix for the mos6502 part of bug #3814. This is a bug that can result in the silent generation of wrong code, and AFAIK at least partly a regression since SDCC 4.4.0. I hope this completes the fix for bug #3814. At least now everything of #3814 that I was able to reproduce using OpenBSD on ppc is fixed. But maybe there are still bits that only get triggered on macOS on ppc. We'll have to wait for results in the farm to see that. * I've checked that mos6502, ucr3k, ucz80, ucgbz80 regression tests pass (didn't rebuild the library though) on OpenBSD macppc. * I've done a clean, configure, and run full regression tests using trunk with the two fixes from next picked on Debian GNU/Linux on amd64. Everything passes. Philipp P.S.: IMO getting these fixes in is worth a bit of delay in the SDCC 4.5.0 release. That way, we can point users to a very stable 4.5.0, when we break stuff throughout this year. |