G. Change Log

Mar 2013 (1.4.1)

Maintaince update. also includes things from the errata page

• Changed from arm-eabi to arm-none-eabi.
• Little html fixes here and there. Thank Glod for directory search-and-replace tools.
• all code: since GCC 4.7 broke my assembly functions, I've recompiled all code with the latest devkitArm (currently 40) for asm compatilibity. The examples and tonclib should all work again. I still have to adjust the text to match though.
• asm.htm: fixed non-matching variable names in data sections's code snippet. Thanks, Gdogg
• gfx.htm: removed a lost semicolon in the blending demo.
• hardware.htm: IO-ram upper limit was given as 0401:03FF, which should be 0400:03FF. Thanks, G M.
• code: Fixed links to grit for m7_demo, m7_ex, tte_demo.
• objaff.htm 11.5: fixed spurious sina and cosa calculations in obj_rotscale_ex() and oac_rotscale(). Thanks, dasi.
• GNU assembler manual has moved to http://sourceware.org/binutils/docs/as/index.html (thanks, Joseph).
• Code snippet at §23.2.1:Basic operations. “x68 asm” should, of course, be “x86 asm” (thanks, Wladimir).
• Some of the memory map entries in §1.3 were … imprecise (thanks Pius).

Oct 2008 (1.4)

… Or maybe not. Silly little errors.

• text: the se_index_fast() function in regbg:map-layout was wrong; the second condition should have used `(bgcnt&BG_REG_64x64) == BG_REG_64x64'. Fixed.
• text: the ldm example in asm:memory did not list the right values for ldmda and ldmdb. Fixed.
• code: removed void* arithmetic in tonc_surface.h (hopefully) and fixed berk.c from the timer demo so that it compiles now. Thanks for noticing, elwing and Ealdor.
• Fixed a few random typos here and there

May 2008 (1.4)

• text: Last batch of spelling/grammar fixes. Thanks guys. Especially Jake.
• text: changed the stuff on #include for less hyperbole and more explanations.

I think that'll be all then.

May 2008 (1.3)

• code: All demos that use text now use TTE to do it. Variable width text just looks prettier than fixed width.
• code: Some of the advanced demos use grit to convert the graphics.
• text: added TTE chapter with information about creating (fast) text renderers for every occasion.
• text: I've revamped the setup chapter. It now covers template makefiles and some potential problems with the installation.
• text/code: the irq chapter and its demo uses the new master ISR.
• text: Fixed the brown text in the pdf. For anyone who has the same problem with CutePDF, go to CutePDF printer->Properties->Paper/Quality->Advanced->Graphic->Image Color Management and make sure ICM Method is not set to Host system. How silly of me not to look there first.
• text: Finally removed the obsolete section for IRQ-handling with older devkits.
• tonclib: There is a new system for text called TTE. It's pretty cool. Read more about it in tte.htm.
• tonclib: New rendering functions. There are now ‘TSurface’ structs defining a rendering surface and basic primitive renderers for different surface types. Functionality includes: pixel, line rectangle renderers, a blitter and floodfill. Key surface types: 16bpp bitmap, 8bpp bitmap and 4bpp tiles.
• tonclib: Color adjustment functions. Fading, blending, brightness and more.
• tonclib: Added tonccpy and toncset, memcpy and memset replacements that actually work for VRAM as well.
• tonclib: Put tonclib documentation online: http://www.coranac.com/man/tonclib/index.htm.
• tonclib: Added tonc_libgba.h, a header with most of the libgba constants and functions names mapped to tonc equivalents.
• tonclib: Changed the master ISR to one that doesn't automatically enable interrupt nesting. It's a bit of a downgrade, but it's probably more appropriate. This shouldn't affect anyone that didn't use nested interrupts. The old version is still available, it's just not the default.

Dec 2007 (1.3b)

• Upgraded the recommendations section with a longer list and examples. If you've read other tutorials then please read this!
• More spelling and grammar fixes (thanks Patater)
• Fix in the template makefile for c++. It's -fno-exceptions, not -fno-expections, you silly boy. (Thanks muff).
• All projects now default to cart-boot, not multiboot. This is partially because multiboot doesn't work in devkitPro r21 (at least, not directly), but also because that's how it's normally done anyway.
• Changed git to grit where appropriate. Also fixed the download links of everything to point to the new site.

Feb 2007 (v1.3b)

As every programmer knows, you're supposed to write down the changes you make while making them. As every programmer also knows, this has a tendency to be forgotten ^_^;;. I probably missed a few things here.

Text:

• There's now a PDF version too, made by CutePDF. It's a nice tool, but apparently sometimes messes up pictures a little. Page-breaks also occur in unfortunate places, but this is a browser problem. It's supposed to be countered by CSS's page-break-inside, but I guess that's not widely supported yet. If anyone knows of a potential fix, let me know. Additionally, if anyone knows of a html→PDF converter that keeps headers for the table of contents, I'd be very interested in that. Mind you, it needs to be able to print a 1.4 MB file, and print it correctly! Some PHP html2pdf tools don't render correctly. Word/OpenOffice are probably no good either, as they have a problem with floating divs and pre tags. Also, Word nearly crashes on reading the file. Hehehehe.
• hardware: GBA pictures and capabillity description. Anyone have a GBA Micro pic I can borrow?
• first: hardware pictures.
• bitmaps: new demo discussion for drawing primitives in modes 3/5. The page flipping demo has been moved forward in the chapter, and the mode 3/4/5 demo moved back to after the data discussion.
• objbg: note and pic on reading tiles as bitmaps, as this this still happens to some occasionally.
• regbg: pic to show what the offset registers really do.
• affine: added inverse 2x2 matrix equation.
• affbg: new structs for affine backgrounds, plus new typedefs and a very nifty method of initializing the affine parameters.
• mode7ex: across the board upgrades and new stuff. It now uses proper graphics, making everything look a lot nicer. New background, new fade, sprite rotation-animation and sorting and different methods of motion.
• asm: the proper form of the chapter is materializing. New structure of sections, beginning with a one on general ASM. More examples and more ways of doing the same thing for comparative purposes. Has a section on common constructs now too.
• New subsection on linear interpolation of luts.
• Chapter indexing. All references are how of the form ‘ch.foo’.
• Some chapters have been renamed. tonctonc is now intro, toncmake is now makefile. Also, luts has been merged into fixed, and the parts on makefiles and editors in setup has been moved to a separate file called edmake. Might move that into makefile too.
• All register and register-like tables now use alternating background colors for easier reading.
• regobj: Different structure for obj_demo discussion.
• regbg: deleted BGINFO stuff, as it was never used much and impractical to boot.
• regbg: New graphics besed on Super Metriod's Brinstar instead of the original Norfair. 's prettier now. Also rearranged stuff.
• affine: updated ‘finishing up’ for new routines.
• dma: discussion of upgraded DMA demo
• interrupts: discussions of the new and (much) improved interrupt handler and its demo.
• Inline functions for fixed-point functionality.
• Every chapter has been checked for spelling and grammar. Again. Sigh.

Code:

• tonclib: All new tonclib, with new file structure. All files are prefixed with tonc so the don't interfere with outside files. The types, memory map and register #defines are centralized in types, memmap and memdef. The main file to include is now tonc.h.
• tonclib: Doxygen comments all around. The resulting documentation can be found in tonclib.chm.
• tonclib: a few of the new items. A brand new interrupt handler for nested, prioritized interrupts. Mode 3/5 line drawers. A new .12f sine LUT with support functions as well as lerping functions. All fixed-point macros are now inlines.
• tonclib: The BGINFO struct and functions are gone. Wasn't worth much anyway. Also removed are internal OAM shadows; it's better that you can define them when needed and can save IWRAM by potentially storing them in EWRAM. All OAM functions now use general object pointers, rather than buffers.
• tonclib: yet another Great Renaming. Among other things: The leading underscore for zero-#defines are gone. I thought it was a good way if guarding against potential unsafe operations, but they just look too weird to use. And there was much rejoicing. Some macros have lots their _ON prefix when it's obvious that that's what they do. OAM structs are now OBJ_ATTR and OBJ_AFFINE and supporting functions are now prefixed obj_ and obj_aff_. BGAFF_EX is now BG_AFFINE and used in most affine BG functions. A complete list can be found in tonc_legacy.h, which you can #include to keep compatibility with older code.

• projects: the structure of the projects hierarchy has been altered. The demos have been categorized as basic, extended or advanced, which correspond with the tonc-text parts. Basic demos are simpler, with simple makefiles. They are completely self-sufficient, which should help learning the ropes. The extended demos have more complete makefiles and make use of tonclib. The advanced demos have devkitPro-like makefiles. As much as I'd like to, the actual DKP templates don't quite suit my purposes (sorry, Dave :P) so I rolled my own. The advanced demos also make use of assembly files for data.

  The project folders also contain .pnproj files, which can be opened and run from Programmer's Notepad.

• projects: New projects. m3_demo, for drawing in mode 3. There are also a couple of new ones in the lab folder. They don't have discussions yet, but they're worth checking out. bigmap should be of particular interest.
• projects: Update projects. All projects have been updated to the new tonclib. The DMA, irq and mode 7 demos have had drastic changes in content. dma_demo is now about using HDMA effects, in this case making a circular window. irq_demo uses the new irq handler to uts fullests with nested interrupts and changing irq priorities. As for mode7ex, well, you'd better just see for yourself.

Jul 23, 2006 (v1.2.4)

• Added a rather long chapter on ARM/THUMB assembly. This is still a draft version, though. Most of the content is there, but I still need to reshuffle sections and spell/grammer check the whole thing.
• And yet more spell fixes >_<.

Jun 3, 2006 (v1.2.3)

• Changed makefiles and build instructions to use devkitARM r19.
• All sections and subsections are now numbered, w00t!
• Added alignment attributes to most structs, as those are now pretty much required if you want struct-copies to work properly. For more, see here

Apr 28, 2006 (v1.2.2)

• Finally realized what caused the 1 pixel offset I've been seeing in affine objects sometimes (thanks NEiM0D). Updated affobj.htm and obj_aff for it.
• Moved the new off-center affine object stuff to its proper place.
• Some small sed usage to convert from GCC error-reports to Visual C++ format, based on this.
• Now that my html auto-numbering system works (at least a first trial), text.htm is now de-reffed. Yay.
• Small changes to interrupts and gfx.
• Added Javascript to make the id attributes visible. Will probably add more later.

Apr 28, 2006 (v1.2.1)

• Apparently no$gba doesn't like it if you use section mirroring, like I did for REG_INTMAIN and REG_IFBIOS. These now use the proper addresses.
• Spelling fixes in intro.htm. Thanks again, Mick.
• New makefiles for extended and advanced projects. It does mean that makefile.htm is now pretty much behind the times.
• Updated setup.htm for devkitARM changes. Changed the figures a little too.
• New chapter called the lab, where I'll place new stuff that's almost, but not quite, ready. Currently contains text for priorities and sprite sorting, and a discussion on affine transformations around a non-center reference point. Both come with new demos called prio_demo and oacombo.
• Added instructions on how to run makefiles via context or PN in setup.htm.
• Added gfx2gba and grit conversion instructions in a few places.
• More notes in bitmaps.htm's data section.

I intend to make a few changes in tonc's code. First, I'll try to decouple the code in the basic demos from tonclib, which should make them easier to understand as you won't have to browse through all the other stuff. Second, this will allow me to rework and optimize tonclib, which is now hampered in some areas by me having to keep a number of things simpler than I'd like to. Now, this is what I'd like to do; I can't really tell when (if) I will get round to it.

Also, I have half a mind of changing the current DMA demo to this one, which simply looks a lot cooler, even though there's is a lot more magic going on. Meh, we'll see.

Mar 21, 2006 (v1.2)

More non-final updates. Quite a lot actually.

• More typos fixed.
• bg_init() never initialized the BGINFO position. Oops.
• Tossed chapter order around a bit. I've moved keys up to right after bitmaps, which is a much better place for it anyway.
• Updated First Demo, Bitmap Modes, Regular sprites, Regular Backgrounds, Affine Sprites, Affine Backgrounds, Graphic effects, and Timers with full or nearly full code of their demos.
• First Demo now has two demos, one purely with hardcode numbers (muwahaha!), and one according to more sound programming principles. Also described these in much more detail.
• Added two demos to Regular Backgrounds, one of which introduces tonclib and its text functions, which will come back a lot lateron. Speaking of which …
• Recoded bld_demo, m7_demo, mos_demo, obj_aff, and tmr_demo to use tonclib's text so that it's clearer what you're changing.
• Replaced copiers/fillers with tonclib's memcpy16/32 and memset16/32 in most demos after its introduction.
• Added field defines to a lot of register tables.
• Restructed part of keys.htm for better explanations of the various functions I use.
• Added a nasty piece on division by reciprocal multiplication. Not for the squeemish.
• Added a simple template makefile and discussion for the second demo.
• Added a subsection on tribool keystates. In fact, nearly all of the demos that might benefit from these have been altered to use them. One line of code instead of four lines, and faster to boot. Seems like a win to me.
• Added a subsection on the proper build procedure, which was still missing from that whole section. This is pretty much required reading for anyone how has been following non-tonc tutorials and adopted their coding standards.
• Merged the fixed-point and LUT chapters, and rewrote most of both.
• jake2431 has been gathering a lot of useful links in this thread: forum:8353. If you're new to C and/or GBA/NDS programming, I recommend you check it out.

Jan 27, 2006 (v1.1)

Heh, so that wasn't not the final update after all :P.

• Added a little note to setup on how to get rid of them useless directories that MSVC 6.0 insists on creating all the time.
• Fixed devkitARM URL and revision. (don't know why I bother with that, though, as there will be a new version the second I post this. Gawddammit, Dave, quit it! >_<)
• More code in the text. At least for the earlier pages.
• Two new chapters: one on Text system fundamentals and producing beeps. The latter isn't quite finished yet, but should be enough to get you going. There are 5 new demos that go with these: 4 for text, one for sound.
• More name changes. This time in demo-names only, though, so don't worry there.
• Added/moved a section on How to deal with data. This explains some of the nastiness you may or may not encounter. But if you do, it's nice to know why they happen and how to fix it, no?

Jun 28, 2005 (v1.0)

Final update. Probably. Not because I'm done with this thing, but because there is so much to change and to add that it's easier to start from scratch again. When I started tonc I still knew very little about GBA programming and tried to do the best I could as I went along. Now I'm a little older and wiser (well, older at any rate), know a lot more about proper procedure, what's useful to have and what isn't and also where people can get stuck on (thank you newbie forum dwellers for all your questions!) From the ideas I'm having, tonc 2 will be a lot bigger, better, and have more explosions! Errr, demos. Tonclib will get a major overhaul with new names and new, optimised functions including text for all modes, memory routines and more. But it'll take a while to get there, so I thought I'd update the original one last time.

• Many, many spelling and grammar fixes. Too many in fact. C'mon people, tell me about these things!
• DMA_SRC_RESET is 0x01800000, not 0x00600000. This is what made the outcome of dma_demo so weird. Also fixed sbb_aff's black cross-hairs, which had its x and y values swapped in OAM. Stupid attribute x,y order.
• Name changes. Lots of them. This partially falls under keeping to GBA community standards (OBJ_ATTR, charblock, screenblock, swi naming) and other classification issues (DCNT_x for REG_DISPCNT and such; prefix underscore for bit-defines that are zero, trust me this is a good thing). I've taken the liberty of creating a legacy.h that redefines all those old names into the newer ones so that you won't have to do all the renaming yourself if you don't want to. The older names are depreciated, though. This renaming is only part of the full tonc2 renaming, but I can't do functions yet because that would break old code.
• Some small functionality changes. Most notably, key_poll() now already inverts REG_KEYINPUT (formerly REG_P1). This is a good thing, because now the synchronous functions will actually make more sense. Also, m4_plot() (formerly _vid_plot8) really does plot per pixel, not per two.
• The memory routines memcpy16/32 and memset16/32 are optimised in assembly, and probably the fastest you'll come across. Rivals the speed of CpuFastSet, but none of the alignment / size requirements.
• swi.s has calls for all BIOS routines. Some extras have been relocated to swi_ex.s
• Added x_BUILD macros for setting bit-flags in clusters. May be useful, maybe not.
• Added rectangle drawers for the bitmap modes. Fairly optimised.
• Fixed list-margins for Firefox. Or rather, fixed list-margins to what the standard requires, but which MSIE doesn't follow. (Now if I can only figure out what to do about that <col> tag)
• Restyled the register tables.
• I finally realised how I could do matrices in pure html instead of images so pretty much all equations are now html. I expect it's quite close to MathML, but since MSIE doesn't support it natively and I don't want to worry you with an extra download (which may not even work on older versions), this'll do for now. (Now if you'll excuse me, I going to lie down to get the feeling in my brain back)
• All sections, equations, tables etc now have id's for linking too and (maybe) automatic numbering if I figure out how.
• int_demo now uses a separate file for the direct isr and makes proper use of sections and ARM/THUMB code. See the demo description for more.

That's about it I think, but it should be enough. I hve bits and pieces of the tonc2 text, examples and lib, though maybe not in their final forms. They are available, but only on request. If anyone has suggestions or requests I'll see what I can do. This also goes for mistakes (the ones that the compiler/linker can't catch) you've made that you think others might make too. I know a good number of them already from the forum (like that you should NOT use bytes or halfwords for local variables, since it can really kill performance, int or u32 only, please!. Pretty please. With sugar on top. And frosting and whipped-cream.) Don't need to know every little thing though, especially if it's already covered by the well-done gbadev forum FAQ or already covered in here somewhere.

If anyone knows how I can keep track of all the header/equation/figure numbering automatically (without CSS2, which isn't properly supported by MSIE :( ) that would be very helpful. Actually, the numbering itself isn't the problem, referencing them is.

Also, I could use more real-life examples of tile-map/sprite collision detection and response. I know the bounding box stuff and the basics of detection (even pixel-perfect), but no matter what I do I seem to be getting stuck on some of the particulars of diagonal movement and when things move at more than one pixel/frame. I'd very much like to see how it's done in real platform games for complex scenes that have multiple sprite-sprite and sprite-background collisions, not just single sprite-bg.

Dec 5, 2004 (v0.99.6)

Added the numbers page about number systems, bits and bit operations. I should warn you that it's rather large. It's been a while since I added something and I think I got a little carried away :\. I may break it up into smaller pieces later. Maybe.

DragonBASIC is in the process of being transferred to a new domain, so the old URL is invalid now. At the moment you can still find the forums here, but the compiler itself is still in limbo for the time being.

Aug 3, 2004 (v0.99.5)

• Made some minor corrections all round. The first mode7 demo and page now use a different name for the camera position, so it won't clash with v from mode7d.
• Added a rudimentary text demo in the form of txt_demo.

I think this will be the last update for a while, for a number of reasons. Firstly, I think I have to actually use some of this stuff, to see what's wrong with it. Secondly, I think I may have to put in some more work into a converter and how to add pure binary files to the demos in a friendly way. Thirdly, PERN's back with a vengeance. As such, there seems little point in developing Tonc any further right now, as it seems that the new PERN is going to be very, very complete.

Jul 16, 2004 (v0.99.5)

• Fixed aff_rotscale2, which should not have shrunk the source-angle, but rather a copy of it. Defined MAPBLOCK to contain 1024 (=32*32) tegels, not 512. Was confused with tile-blocks.
• Made a lot of small changes to tiles/map functionality. All map/tile structures are now simple typedefs so you can access their internals via a simple array-access rather than (inline) functions. The inline functions themselves have been removed.
• Changed then BGINFO struct a bit, and added some map functions.

I'm sorry if any of these changes causes you any inconvenience, but I think it's better in the long run.

Jul 11, 2004 (v0.99.4)

• Deprioritized MSVC makefile projects in setup.
• Mopped up several minor and not so minor errors in mode7ex. Well, I did say there'd be some I hadn't discovered yet.
• Renamed the interrupt requests for registers REG_DISPSTAT from X_INT to X_IRQ, which is more proper.
• Modified swi.htm to show how to use pure assembly for this purpose and added a small section on the aapcs. Also renamed some of the affine structs and functions; you have been warned.
• And yet more typo fixes, where the h311 do they keep coming from? I swear if I find one more "it's"/"its" mixup I'm going to scream. [Later that day] Right, that's it: AAAAAAAARRRRHRHRHRRGGGHHHH!!!
• Fixed a number of rot-scale equations that had the rotation and scaling ops wrong in the intermediate steps. Oops.
• Fixed a window control macros (forgot some shifts). Should work properly now. Should.
• Added geom.h|.c to the library, as I intend to use points and rectangles more often. Also added ABS, SGN and SWAP macros.
• All multiboot demos (i.e., all of them) now have the extension mb.gba to indicate them as such.
• Renamed key_pressed() to key_hit(), which should cause less confusion about what the function actually does (thank's^H^Hs for the name Dark Angel (see? The apostrophe occurs almost automatically :( ).

I'm working on a nice text system right now. If anyone has any requests I'll see what I can do.

June 27, 2004 (v0.99.3)

Ahhh, home at last, where I have a proper computer and Kink-FM blasting through my stereo, excccellent! =)

• Added -Map and -Wl command-line options to the flags list.
• Moved the graphics data that is only used once into the demo-folder where they are used; the gfx directory now only has shared graphics in it.
• Had to write swi.s again because the `utils clean' command destroys all .s files if there is a .c with the same title. Make sure that there is no swi.c in your utils directory when copying the the new stuff.
• The mode7ex.htm page is finally complete. Yes I know it's long and full of nasty linear algebra; if you have trouble getting through it and/or have suggestions on making it more readable, plz, do tell.
• The accompanying mode7d demo is just about where I wanted it.Sure there are still some minor problems, but it should be enough to get you started.

June 21, 2004 (v0.99.2)

• I made a lot of changes to mode7d; all the real mode 7 code is now in separate files so using it in other projects is easier now. Though mode7ex.htm still needs a lot of work, you can find most of the text in draft-form in m7theory.zip. Yes, it's a Word document; yes, I know that sucks; yes, I will convert it to html when I the text is stable and understandable (please tell me what I need to change in this respect); and yes, I will do this conversion manually, since Word should be allowed to approach HTML to within 500 yards. Perhaps more.
• Made some minor fixes to the matrix page. Silly me, I got the cross-product definition all wrong.
• Added info on REG_P1CNT to the keypad page. Yet another thing which only this site covers :).

Devving on a P2-300 with 24MB RAM: VBA runs at 50% (and 23% for mode7d) and minimizing a window takes a few seconds. Man, this sucks.

June 11, 2004 (v0.99.1)

• Fixed the style sheets so that the background image, colors and borders, etc, etc, appear as I had intented on Mozilla. Sorry about that, didn't know the the wrong comments would screw it up so much. Made a vallidation run and got rid of all nonvalidities ... except one: the <nobr> tags that I need to keep certain things together.
• All BIOS calls are now inside swi.s, in assembly. Which is where they belongs, really.

June 3, 2004 (v0.99)

• Found out about the wrapping artifact, and changed the sprite pages accordingly. obj_aff now allows moving the sprite so you can see this artifact for yourself.
• Finally got over my dislike of near-empty directories (the files get so lonely that way) and put all the demo-code in separate directories. Now, if only I could get over my ifphobia as well...
• Added a section on vsyncing with interrupts in swi.htm and an accompanying demo, swi_vsync. You need to see these.
• Added int_enable_ex and int_disable_ex, which should make working with interrupts easier. However, I am not 100% sure if I got all the registers and flags right.
• C++ doesn't like it when you try to use a struct-copy on a volatile variable, like bga_update_ex does. Or did, I should say.
• Learned some new CSS tricks and am busy updating and structuring the layout of all pages. It's mostly subtle stuff though, like standardizing the equation layout and giving code and register listings a subtle border that makes it stand out in printing. Non-subtle is that ever image should have a caption now.
• Resumed work on mode7ex.htm and its accompanying demo, mode7d. Adding variable pitch turned out to be easier than I thought, w00t! It's still a little buggy, though.

With all these changes, it is adviced to save or remove older Tonc stuff when upgrading to avoid double files and other inconsistencies.

May 24, 2004 (v0.98.5)

• devkitARM is now the primary devkit for Tonc. Makefiles and text are updated to match the change.
• Using a separate interrupt file now instead of a custom crt0.S and got The Point® of the critter in the process. The text is modified to reflect newer insights, as is int_demo.c
• Started work on a glossary.
• Added instructions on how to run makefiles without Visual C++ in the the Tonc-code readme. Silly that I never thought of that before.
• REG_IF is at 0400:0202, not 0400:0200, doh!
• As you can see, I'm trying to use context-specific bullets for log entries. I still need to figure out what images to use for what purpose, though.
• Rewritten the build_all and clean_all targets in tonc.mak.They're quite nasty now, but act more correctly and allow me to switch to a "one demo, one dir" structure if people finds having everything in the examples folder a bit messy.

I am soooo tired right now so I wouldn't be surprised if I messed up somewhere. I'll get it fixed when I've had a chance to sssslssszzzzzzz....

May 16, 2004 (v0.98)

• Lot's of changes. First of all, I finally have a means to test on hardware, Wheeeee!! However, it did point out that you can't use the object tileblocks for backgrounds after all :(. Added my early experiences with hardware tests on a number of the pages.
• Including in the windowing section. It seems that you need to be really careful with the vertical settings of windowing. Updated the windowing demo to not use u16 arithmetic for the window size (which is given in bytes), and more precise movement.
• Threw the DMA code around. I'm now using dma_memcpy() for general copies, and renamed the old DMA_CPY() macro to DMA_TRANSFER, and rearranged the order of arguments to match memcpy. Makes more sense that way.
• Also changed oi_set to oi_set_attr and oi_pos to oi_set_pos.
• The Tonc Utils configuration now compiles the utility code into a library. Required for mode7d.
• Created a build_all rule in tonc.mak. Rebuilding everything by hand was really getting on my nerves.
• More random clean-ups. The images on the entrance page are now links as well. Important notes are now in red boxes, for extra visibility. Added the demo-code of bm_mode.c, to show the basic steps of loading a picture and using keys. I should still post full code that earlier in the tutorials. Added an example of a fixed point identity matrix on affine.htm to make sure people don't try to use floats.
• Fixed swi.htm. Again! I swear, if I find one more error here some somebody is gonna get hurt. And it won't be me. This time the range of arctan2 was wrong (should be full circle). What makes this error even worse is that I should (and did) know it's supposed to be the full-circle all along; it is the raison d'être of arctan2 after all.
• Matrix page is done.
• Practically fell out of my chair laughing at villainsupply.com. Ouch.
• Nearly drowned in my own drool after watching the Nintendo stuff at E3. Gargle.
• The picture used in key_demo had the palette-indices of KEY_START and KEY_SELECT switched. I never really noticed because emulators don't have real start and select buttons. So once again, hardware testing saves the day.

Apr 29, 2004 (v0.97)

• Added links to devkitARM as well as instructions on how to get it working. There is a very real possibility that I'll switch to this toolchain in the future.
• Converted most of the macros to inline functions. Safer, easier to read and just as fast. Yes, plz.
• Some more diagrams about tile-counting and the affine transformations.
• Still to do: finishing the matrix page (and perhaps the mode7ex page). And now that I have my tile-map editor Mirach, I may be able to do something with that as well. And I really, really need to get working on a text-system.

Apr 9, 2004 (v0.97)

The object affine functions have a background counterpart now, and mode7d is coming along nicely.

Apr 4, 2004 (v0.96)

Renamed OAM structs and related items. Again. When is this gonna be final?!? Also, thanks to Lupin's problems with sprite-placement in 3D I finally got my matrix-transform sense back. Now that I get it again, I hope to expand the mode7 section in the near future. I already got a working example for 3d-sprite placement already in the form of mode7d.

Mar 31, 2004 (v0.96)

Working on a vector/matrix page, some reshuffling of page-order and yet more random little cleanups.

Mar 24, 2004 (v0.96)

Replaced the assembly listing in swi.htm with the proper THUMB listing. I'd forgotten I wasn't using ARM code anymore.

Mar 20, 2004 (v0.96)

And just when you think you're finished you find another 2 things you can improve upon. Argh. Anyway, I've changed the way sine and cosine are retrieved. They're both macros now, using one 512-entry long s16 sine-LUT. Also, I finally realized how to XOR the vid_page directly for page-flipping. And, oh yeah, the compiler flag for compile, but not assemble should be -S, not -s. Oops.
I think I finally know how I can modify my affine functions to apply to backgrounds without having to use the OBJ_AFFINE structure, but it may be a while before I actually do that.

Mar 17, 2004 (v0.95)

Added ArcTan2 function to swi_demo and fixed the errors that swi.htm still contained. Argh, and I thought I'd been thorough in weeding out all inconsistencies after the recent name and code modifications.

Mar 14, 2004 (v0.95)

First entry in the log. I've rewritten the parts about sprites and backgrounds, changed glyphs to tiles and tiles to tegels (hope I got them all :]), updated all the code one last time and written sections on how to set up DKA, MSVC and makefiles. I think Tonc's ready for use now, wheee!