[uae] Re: uae/winuae/whateveruae (was Re: No mouse in fullscreen)
- From: Toni Wilen <twilen@xxxxxxxxxx>
- To: uae@xxxxxxxxxxxxx
- Date: Thu, 25 Jun 2009 20:15:00 +0300
Thanks for finally saying something I can understand :)
When I say "crap" I primarily mean changes that don't change
functionality, but make the diffs huge and hard to read for no good
reason at all:
* unnecessary whitespace and formatting changes
In my opinion current formatting totally sucks (either tabs or spaces,
not both, perhaps this is some "*nix braindamage..") This is something I
will change sooner or later.
* things like the patch that adds a "regs" argument to all CPU
functions
Blame E-UAE author. I tried to get rid of it but it only broke stuff and
I didn't bother with it anymore.
* unnecessary added casts in lots of places
That was attempt to make it C++ compatible. It didn't work. It would
have made many things much easier.. (afaik this should be done someday
before all APIs/GUIs/whatever are C++ only and we need wrappers
everywhere..) Note: do not think about *nix only where C is much more
common (afaik)
* Code mangling for Windows Unicode braindamage, crapping non-portable
Windows stuff all over the supposedly portable core code.
There "should not" be anyhing non-portable, I could have made it very
unportable if I wanted. Did you even check it? Type/functions names may
be Windows specific but it is not unicode-only, it still works just fine
with ascii or even as utf-8 in *nix if needed. Simple search and replace
can fix this if there really is fully portable namespace. (if there
isn't, again, too bad because I ignore "lowest common denominator"
reasons, was burned once too many times)
This was the best solution I could find that won't break portability.
(is there even portable solution? if not, "too bad")
100% internalization support is basic feature today, at least in Windows
world. ("and I thought I was alone")
This kind of thing is just inconsiderate, unsurprisingly when your own
stated goal is to do what you want without the bother of working with
others. Since you also don't send me patches with explanations of
what's being fixed, I've occasionally tried to go through digging
expeditions through the diffs to find usable changes, but since it's so
hard because of things like the above, it invariably leads to burnout
after a few days.
Again, you never said you wanted patches or even worked with UAE
anymore. (and I can almost guarantee you would be the only one wanting
patches)
And because I just "hack stuff", I don't know if it fixes something
until much later. Unfortunately this is how I do this stuff. Difficult
to do it other way because there are still unknown and odd features in
Amiga chipset. (like vpos counter gets increased at hpos ~3-4, NOT at 0,
this is 100% confirmed with logic analyzer)
Some of the new features aren't problematic, if something lives in its
own file like akiko.c and doesn't add a lot of intrusive code elsewhere,
I don't mind. What I object to is new features implemented in the
quickest way possible without consideration for portability (e.g. the
incredibly awful AHI stuff, which totally ignored the existing sound
abstractions). Also, new features that expose a user-visible interface
AHI has nothing to do with Paula audio, except perhaps mixing. AHI can
be 8/16/32 bit. 2 or 7.1 channels. Can have rates much higher than Paula
can generate etc. It does not need any Paula specialties like DMA
sequencing etc, AUDxDAT etc. Paula is the weird one. AHI is very simple,
no side-effects.
I didn't do the win32 AHI code but I'd still would have separated them.
Paula audio is already too complex (if you want to emulate everything
exactly) (I do agree current AHI code is weird..)
(e.g. in the config files) that isn't thought out properly and precludes
any further improvement (the inputdevice stuff comes to mind, which I
find way overdone with a terrible interface).
Do you mean the GUI? It is crap but that can be improved easily later.
Inputdevice may be overdone but it allows anything. Inputdevice
configuration _is_ crappy and overdone too (single configuration is
probably more than enough, no need for 4..)
Most importantly it does work and does exactly what I wanted it to do
(=everything)
We ("we" what?) are not developing something like Linux Kernel that
needs high coding standards, especially if there is only something like
max 3 people in the whole world that really knows UAE and Amiga hardware
inside out..
It's not a small toy either. I vehemently disagree with the notion "we
don't need high coding standards" for a project of this size, and I
think this one of the core problems why we can't work together. Yours
is an approach that only works in one-person projects.
In my mind it has been one-person project for last 2 years at least.
Nobody said anything, I decided nobody is interested.
You have good reasons to be annoyed but I really thought you didn't care
anymore (nobody else did either) I was totally alone and I learned to
work alone. It may be too late now..
Another problem seems to be Windows<>*nix world "collision". So many
differences.. I can hate Microsoft like anyone else (they do lots of
stupid things too) but Windows does now have great "multimedia" APIs,
better than most other operating systems.
I hope we can keep OS issues separate from real technical issues.
I can try to forget my bad habits from last few years but it won't be easy.
Other related posts:
