Re: [hatari-devel] OT: My dream about Hatari

[ Thread Index | Date Index | More Archives ]

Le 16/07/2013 14:21, Miro Kropáček a écrit :
Hi guys,

sorry for the poetic subject but basically, it's true :) It's been some
time when I had realized that if I want to code something serious (like
an "Amiga style" demo), there's no way I can do it in 100% assembler
with Devpac+qdsp+dspdebug threesome, i.e. that I need to organize
structures in C(++), make it possible to debug in gdb (ideally on host)


So this is my last appeal to you, please implement a debugger into
Hatari, a real debugger with stepping, changing values (regs, memory),
UI (mainly UI ;), automatic symbol names (it's not there yet but I plan
to import them from .LOD files, it's super easy) ... it really takes
only two weeks and it would help me a lot (because my debugger has
several limitations of course, no interrupts for example) and I'm sure
not only me.

Sorry for such a long post but I just wanted to show you that something
which might seems as a b***h work can be actually pretty simple to do.


your work is interesting, but I'm not sure anyone is denying that a new "modern" UI that fits better with today's OS would be better.

I'm a big fan of Steem debugger and it's flexibilty on the number of extra windows you can open with different contexts, breakpoints and more generally all the things you can do from its UI.

Now, back to Hatari, the people developping it just lack time and/or interest to work on UI parts.

In my case, I'm mainly interested in accuracy and low level hardware emulation, that is reaching a point where the emulator could really match the real machine at the bus level for example ; honestly, coding UI is really on the bottom of my todo list, not that I deny it's not required to improve it, but that's not what I plan to do as "fun" on the spare time I can devote to Hatari.

At the moment, I don't think I can spend more than 2 hours per week on Hatari, so assuming the UI you wrote took 2 weeks, I guess it would have required me 3 months or more of spare time to do this "2 weeks/man" work. And starting a new UI from scratch is not something you can do by assuming it will take 6 months from now before reaching a result, due to lack of spare time. It's really not motivating.

But there're other options : we already have some gui-sdl and gui-osx directories, so next logical step would be to have a gui-qt directory for example. As already discussed in the past, this would require some rework in Hatari (add an abstraction level for different UI, linking Hatari C code with QT C++, ...), but nothing that can't be done.

Given the work you made to build your UI and the fact that you seem to be interested by improving the UI for demos' developpers, maybe you could be the man to work on this part in Hatari ? Emulation's knowledge is not necessarily required, it's more about handling windows and mapping some data structures in a more user friendly layout.



Mail converted by MHonArc 2.6.19+