RAM to do so, which I agonized over for about a week, but I feel it's
a fairly good tradeoff for the features we get. Eventually, I may come
up with a more accurate phosphor mode which eliminates this approach
entirely.
Reverted OpenGL framebuffer to 32-bit colour mode. We're using Blargg
effects, so we may as well get the full fidelity out of it. Besides,
testing has shown that there's not much speed difference between 16
and 32 bit modes, making me think that 16-bit is probably be swizzled
to 32-bit anyway.
Bumped release date to June 1, version to rc1.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2501 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
we need to modify what is caught. Ironically enough, this fixes
a bug whereby when certain errors occur, the recovery system
itself would crash, not the issue that actually caused the bug!
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2499 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
before the release tomorrow (since I'm not entirely sure how it
works), but I'm going to give it the old college try.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2497 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
for phosphor effects. It's just not ready to go yet, more research is
needed, and I just don't have the time to fix it. Besides, it's been
two months since the last release, and I need a vacation.
Because of this, the release date is changed to May 25.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2494 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
instead of the entire area. This fixes the speed regression from 3.6,
since the main TIA texture is now larger than before. Previously it
was 160 pixels wide out of 256 total (rounded to power of two), now
it's 560 wide out of 1024 total. In 3.6 it was updating a buffer 256
pixels wide, in the latest beta releases it was sending 4 times as
much data (256 -> 1024). Now it sends only 160/560. Similar
improvements happened with the height as well (210 lines vs 256).
On my test systems at least, this is an extremely large win,
resulting in similar speeds as 3.6 even though we've moved to a 32-bit
framebuffer.
Bumped version # for next beta release.
Pushed release date to June 1, to make absolutely sure I have
everything fixed and ready to go.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2493 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
why compiling for multiple platforms with different compilers
really helps the robustness of code ...
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2492 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
The documentation states that this command is aliased, the current
frequency should be returned instead of $F2; this part isn't done yet,
since I'm still working on actually calculating the frequency.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2491 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
source code header file. As you can see, this is one of the more
complex schemes (and it isn't even finished yet!).
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2490 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
blending in conjunction with phosphor effect.
For the Chetiry scheme, when loading or saving the 64 byte score
table, we now ignore the first 4 bytes (as specified by the
documentation). These 4 bytes/addresses are actually a type of hotspot.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2489 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
now load and play test ROMs, complete with emulation of timing delays
due to slow accesses on real Harmony hardware. Still TODO is the tunes,
DPC+ stuff, which is stubbed out at this point.
Fixed bug in EFSC bankswitching; state files didn't contain extended RAM
information.
Cleaned up the Serializer API, resulting in slightly faster operation
and smaller state files. Because of this, the state file format
has changed for this release (old state files will no longer work).
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2487 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
cases, these archives contain directories named '__MACOSX' which
contain link files that were being erroneously detected as actual
ROMs.
Added FilesystemNode::isFile() method, and updated several places
in the code to use it. Previously, determining whether something
was a file was simply testing if it wasn't a directory, but this
logic isn't always valid (it's possible to be neither a regular
file *nor* a directory).
Bumped version # for RC release. It's getting very close ...
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2477 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
- Alt-9 / Shift-Alt-9 selects (increases/decreases) between
the various adjustables
- Alt-0 / Shift-Alt-0 increases/decreases the actual values for
the previously selected adjustable
Cleaned up the API a little, moving some stuff from Console to
FrameBuffer class. I think we're due for a pretty major
reorganization of parts of the codebase very soon.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2473 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
Settings::loadConfig() and Settings::saveConfig() when required.
This fixes an issue with OSX in particular, or any port where
those methods are overrided.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2471 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
the Video Settings UI. Still TODO is add key shortcuts for the various
adjustable TV effects (contrast, brightness, etc).
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2468 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
'tv_scaninter' option. Added Alt-8 keypress to toggle this; still
TODO is add a UI item for it.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2462 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
below the base directory weren't being detected. The ROMs can
now be in any directory arrangement (although they'll still be
presented in the UI as if they're all in the base directory).
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2460 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
26-27% to 11-12% on my test system. This is from a baseline of 8-9%
for normal rendering, so the TV effects are now quite fast (only an
extra 2-4% CPU usage).
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2459 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
left to do, but the basic functionality is now there. Alt 1 - Alt 6
selects among TV modes as follows: off, composite, svideo, rgb, bad and
custom. The latter is just composite at this point; there's no code to
actually change specific adjustables yet. Alt 7/Shift-Alt 7 increases/
decreases scanline blend intensity (20% - 100%).
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2456 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
the patches I received in the past, but also includes stuff from newer
versions. It compiles correctly; now I just have to activate it and
pass it a framebuffer and see what happens ...
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2454 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
basically the skeleton of F4SC, since it's partly modeled after that
(8 4K banks, etc). More work will come after the 3.7 release.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2452 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
implementation ready by the end of the coming week. First, I have to
convert a patch against Stella 2.8.4 and an old Blargg version to the
current codebase (and a newer Blargg version).
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2451 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
instead of having different keys to turn them on and off separately.
This brings the functionality in line with the rest of the objects/
collision key-combos, which were also toggle-only.
git-svn-id: svn://svn.code.sf.net/p/stella/code/trunk@2449 8b62c5a3-ac7e-4cc8-8f21-d9a121418aba
stephena