I just added a module to my gbz80 project which uses floating point. Turns out that this adds way too much code. Investigating the library source indicates that the situation would be helped by the inclusion of a floating point library implemented in assembler.
Is there such a library around here? Are there any plans to include one?
Philipp Klaus Krause
There currently are no plans to include a fp library written in asm for the gbz80 or z80.
I can see that code size is a problem on the gbz80. As maintainer of the Z80 port, and I have worked towards code size reduction on the z80 with some success. A few of those also affect the Z80. It probably is possible to port most of the others to the gbz80 port without too much effort. Unfortunately the gbz80 emulator is broken on 64-bit systems (bug #1929317), so I cannot just try porting by trial-and error using the regression tests.
Philipp, are you sure that the gbz80 emulator is broken only on 64-bit systems? This is not mentioned in the bug #1929317 report.
And what it means "broken on 64-bit systems": running on 64-bit system or compiled for 64-bit system?
In the later case we could compile the emulator for 32-bit system using "-march=i386 -m32" gcc command line options, at least as workaround until the proper solution is found.