icon.library 46.4 test versions

[Foul]

happy to help

[PeterK]

Ok thanks Foul.

Just curious, is there any speed gain noticeable on your system?

This def_icon bug should be fixed in 46.4.245 now.

[Foul]

Without blitzkick and custom graphics.library, my system is much slower.
I dont need any bench to show that, my eyes are enough.

Just opening big games drawer could be twice speed :-)

edit : 46.4.245 seems ok...

[PeterK]

Updated to icon.library 46.4.246

Fixed a small bug concerning app-icons.

[Bamiga2002]

46.4.246 runnin' good. What I marked was that icons seem to display a tad faster when disabling PowerIcons-patch.

[PeterK]

icons seem to display a tad faster when disabling PowerIcons-patch

Yes correct. There are no effects, no speed increase, when the PowerIcons patch is running, because PowerIcons needs the colormapping of the selected image for the icon dragging and that was already reduced to a limit of only 40 colors since the old days of IconBeFast. But without PowerIcons there is just one image now which has to be mapped to the screens colors instead of two.

[Foul]

i'm learning asm and when i'm trying to assemble sources, i got this error with Asmpro 1.18 :



it seems to refer to "Icona_Dummy", but nothing found... is it normal ?

[PeterK]

Hi Foul,
these declarations can be found in the Commodore includes, in icon.i for this case.
You have to tell your assembler which files you want to include for the compilation and where your INCLUDE: directory can be found. All the included files that I used are listed at the top of my source code. I'm using PhxAss as assembler with 68020 support enabled and all symbols are case sensitive. You may not find some of the LVOs, because I made my on file for these symbols.

Do you have the NDK already? http://os.amigaworld.de/index.php?lang=en&page=12

[Foul]

all includes installed and found during compilation...

will check NDK and icon.i,

thx

edit: my icon.i and lvos.i are different .. maybe because i'm using 3.1 ones...

[Foul]

again sorry :



After that, i stop annoying you with asm things

[DrCinicus]

Use move.l instead of mevea.l

Bye
Fabio

[PeterK]

@Foul & DrCinicus
Thanks for reporting and correcting this wrong instruction, but PhxAss ignores and corrects that typo anyway.

Updated to icon.library 46.4.247

Fixed two small bugs and a big one which is called blindness

[PeterK]

Updated to icon.library 46.4.248

Just another small fix for a bug introduced in 236.

[altcomputing]

Hmm, I found that when I degrade my AfAOS+PNG Icons system to boot to PAL, icons that will otherwise revert to default images, display trashed (with or without AfAOS running). But If Picasso96 is initialized (but not used to display icons) they are OK in PAL modes.

[PeterK]

Hello altcomputing,

I guess, that happens because IPrefs sets the default icon memory to FastMem on your system, which works correctly with the Picasso drivers, but not with a standard PAL configuration. If you want to use a pure PAL configuration then you should either install FBlit on your system in order to use FastMem for icons or you have to open the Workbench preferences and set the icon memory type back to ChipMem.

[altcomputing]

Originally Posted by PeterK

Hello altcomputing,

I guess, that happens because IPrefs sets the default icon memory to FastMem on your system, which works correctly with the Picasso drivers, but not with a standard PAL configuration. If you want to use a pure PAL configuration then you should either install FBlit on your system in order to use FastMem for icons or you have to open the Workbench preferences and set the icon memory type back to ChipMem.

Tried that... no success.

The thing is that when WB is on a PAL screen, but Voodoo driver is present in DEVS:Monitors, and has therefore been initialized, icons are OK on a PAL screen. But If i just remove Voodoo driver and reboot there's a mess on the screen. So it seems to have something in common with Picasso96.

I think that icon.library guesses from the fact that the RTG is running to move icon data to Fast, regardless of what is actually being used to display WB screen... But I'm a no programmer, just my guess.

I will try to focus on it and give more details.

[PeterK]

I think that icon.library guesses from the fact that the RTG is running to move icon data to Fast...

No, it's IPrefs v44+, that is reading ENV:sys/Workbench.prefs to decide if you want FastMem or ChipMem for your icons.

My icon.library (the most recent versions) just checks if FBlit is running. And if so, then the icons will get FastMem too.

[altcomputing]

Hmm... I'll have a closer look at it then. Maybe Workbench Prefs Editor is fooling IPrefs... Will try to delete prefs and give it a try. But anyway, why does it work when I remove RTG driver?

[PeterK]

I've just tried out a pure PAL system without P96 monitor drivers installed and it works here, no matter if I use FastMem or ChipMem or if FBlit is installed or not.

Please, send a screenshot and some more details about installed libraries.

[altcomputing]

Grab1.png is the issue - There's a regular OS3.9 BB2 + some updated libs, LoadMonDrvs was not run in startup-sequence, therefore it defaults to PAL:HiRes.

Grab2.png is exactly the same, but with LoadMonDrvs ran at startup. The Voodoo driver has been initialized, but not in use - still a PAL screen.

I also tried leaving out the Voodoo driver from DEVS:Monitors and trying that way with LoadMonDrvs so it only loaded standard monitor drivers - result is exact to Grab1.png.

Forgot to mention, those messed up icons are PNG ones!

[PeterK]

Which version of the graphics.library is installed and are there any patches?

[altcomputing]

graphics.library is regular OS3.1 one, patches are out in this config, even MCP is not running.

[Foul]

Setpatch version ?

[altcomputing]

SetPatch is 44.38, AmigaOS ROM Update from 3.9BB2 and scsi.device 43.45.

[PeterK]

Please try this ADF-image, it's important to have mathtrans.library and mathieeedoubbas.library installed. This ADF has all files to display all icons successfully. Just compare the contents with your system.

[altcomputing]

First I checked on my "Low-end" setup:

With LoadMonDrvs (and Voodoo init) - the Qt, Prefs, Shell and CED icons are correct but lowcolor images, rest have fallen back to default "hammer" images.

Without LoadMonDrvs - the Qt, Prefs, Shell and CED icons are correct lowcolor images, rest are trash as usually.

But it works perfect from your bootdisk (it even liked my 060).

Gonna take a look inside and see what you have in there. May be some library or whatever...

[altcomputing]

I use HSMathLibs, but i replaced them with your libs. workbench.library is same, icon.library I updated from your disk. Everything else seems to be the same except for datatypes.library which i have version 44.47. No luck still.

Running FBlit solves the problem on my system, but why? I have to investigate that!

[altcomputing]

Hey, wait a minute, why on your bootdisk all icons are PNGs, but on my system they fallback to default when on PAL screen?

EDIT: Got it, infoPNG was missing. A Pole would say: It's a different pair of shoes now

[Foul]

aargghh still no way compiling asm sources.. even with PhxAss 4.44 or AsmPro :\

All the rest working well

[StingRay]

I just had a quick look, if you want to assemble the source with AsmPro you'll have to change all dbra.s/dbra.w instructions to just "dbra". Dbf instructions don't have any distance specifier so dbra.s/.w is illegal.

[Foul]

with or without modifying all dbra*, still got this error :



Same error with PhxAss 4.44

[StingRay]

Rename the DrawInfo label below to _DrawInfo and change beq.s DrawInfo to beq.s _DrawInfo.

[Foul]

_Drawinfo renamed, ok, another _Screen renamed but :



Not very easy to learn asm ...

[StingRay]

Probably the wrong source to learn Asm. Rename ifdef to ifd.

[Foul]

working but still somes errors... i stop here ... thx anyway

[StingRay]

You're welcome. =) I may have a closer look at the source soon and fix it for AsmPro. Maybe.

[Foul]

it's not very important but, i can learn some tricks using differents assembler... got AsmPro (my favorite), AsmOne, PhxAss, Devpac, and more...

Thank you for your help !

Time to go back to Peter's Topik

[PeterK]

Originally Posted by altcomputing

I use HSMathLibs, but i replaced them with your libs...... Got it, infoPNG was missing. A Pole would say: It's a different pair of shoes now

Congratulations altcomputing,
you can keep using the HSMathLibs, and you can improve the speed of the png.datatype a little by replacing the 68000 version with the 68060 png.datatype. http://aminet.net/package/util/dtype/PNGdt44
Please, also have a look into the readme of my Aminet release for other optional libraries.

@Foul
Compiling my icon.library is NOT offtopic here. So, please don't hesitate to ask anything you want to know about compiling my messy sources. I'm using PhxAssGUI with the following settings (but PhxAssGUI won't work on the normal WinUAE, since it's a little buggy, tries to read from non-existing memory at -4)

@StingRay Thanks a lot for your help and suggestions! Indeed, you're right, DBcc always has a 16 bit displacement. I will change that. The problem with DrawInfo can probably be solved with the PhxAss option "no extern references". The sources can be compiled without any changes if you use my settings below:

[Foul]

thx you very much !

i'm using fs-uae and will try now with your settings...

still got errors with yours settings...

PhxAss : 4.44
PHXAss-GUI : V2.1

don't know why but PHXAss-Gui does't keep all settings when quitting... must re-configure each time @ launch...

[PeterK]

don't know why but PHXAss-Gui does't keep all settings when quitting... must re-configure each time @ launch...

That's the effect of this bug in PhxAssGUI:

...but PhxAssGUI won't work on the normal WinUAE, since it's a little buggy, tries to read from non-existing memory at -4

It needs a particular result when reading from non-existing memory at -4. Usually, on a standard Amiga you will get 0xFFFFFFFF or -1 back, I guess. But UAE will return zero instead, except for my self-compiled WinUAE 1.11. Just in case, that I remember these things wrong, it could also be the opposite of what I said about the result.