Re: [hatari-devel] Release time for 1.8.1 ?

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

On torstai 07 toukokuu 2015, Nicolas Pomarède wrote:
> Le 07/05/2015 00:56, Eero Tamminen a écrit :
> > Some thing that I'd like to see finalized before release:
> > - mouse warp
> > - fopen trace flag
> > - emulation halt dialog in WinUAE CPU core
> > - Douglas' DSP debugger & FPU dissembly issues [1]
> > - Laurent's asm56000.ttp / GEMDOS HD emu issue
> > - non-ASCII SDL GUI / filename support
> > - update Python GUI for new features
> > 
> > [1] Douglas, now that you got latest Hatari working, could
> > 
> >      you recheck these issues whether they still happen?
> > 
> > More testing is probably required for:
> > - SDL2 version
> > - non-ASCII filename support in GEMDOS HD emulation
> > - GEMDOS HD emu Pexec7 stuff
> > - Your recent MIDI changes

Anybody on the list can do testing (only SDL2 item
needs specific build), it doesn't need to be Hatari
developers.

For pexec7, it's enough just to try running different
programs and see whether there are regressions, for MIDI,
running different MIDI programs and seeing whether
there are regressions/problems.

For filename support, it's enough to try Hatari on
directory with weird file names, and see whether programs
(e.g. replacement file selectors) behave with them as expected.


> > And I would really like something to be done for the zero-cycles
> > regression with WinUAE CPU core, especially if this is going to
> > be .1 release instead of .0 one.  It makes profiler much
> > less useful with the new WinUAE CPU core version.
> > 
> > -> could you look at it e.g. with Laurent?
> 
> can you point to an example / mail where this was discussed, I don't
> remember when it was.

Check the early 2013 mails about "zero cycles" and end of 2014 mails
about "profiler cycles".  You participated in latter mail thread. :-)

If you set DEBUG define in src/debug/profilecpu.c to "1", and
enable profiling in debugger with "profile on" command, profiler
will report all cases where there are zero cycles for 2 successive
instructions, according to CyclesGlobalClockCounter counter.

There are *a lot* of them, just from TOS v4 boot.


> > I have also one profiler feature (NatFeats API to temporarily
> > suspend profiling) that's waiting Douglas to have time to test it.
> > 
> > All this might take more than month.
> 
> I agree, but are we sure this will be done in a fixed even after one
> month ? At one point, we need to try to set a fixed date, we can
> postpone a little, but if there's not enough manpower/time to handle
> your list, then we will release anyway (on the opposite, I know I will
> have time to do a release in one month, but after that it's less than
> certain, could be delayed to much later)

OK.  I'll try to work on them next week.


	- Eero