Let's make a nice workbench

Ah, thanks @PeterK, that makes sense (and seems reasonable considering I've not heard much about this before).

I've also been reading that FBlit/FText and P96/CGX is a no-no. Granted, I want to build something for a bog standard A4000, but this would create some extra steps if one were to install an RTG card. So, I'll skip FBlit/FText for now.

Ok, so for performance, working plan is:
- PFS3 file system
- OS3.9 BB1+2
- PeterK icon.library
- StackAttack
- Loadmodule - is this valuable, or problematic?
- Are there any 3rd party libs worth installing? BB3+4 have been problematic for me in the past so unless there is some compelling reason, I'll skip them

[PeterK]

FBlit OR P96/CGX is highly recommended on all 68020+ systems to make it possible to load icons into FastMem or graphic card memory. Using ChipMem is terrible slow.

Yeah, understood. But these items conflict. Maybe I can set it up so that FBlit automatically shuts itself off if P96 or CGX is installed. Should be easy to set a conditional test in S-S.

[Minuous]

@esc:

>BB3+4 have been problematic for me in the past so unless there is some compelling reason, I'll skip them

I suspect that was probably a considerable time ago, do you remember which version? The current version is 1.36. Most older versions of software have bugs, that is no reason not to use a newer version where such issues have been fixed. BB3&4 is really quite a conservative set of changes, especially if only the core package is installed. Any component that causes problems is fixed or removed. For the past year or so there has been no need for any fixes/removals as it has been quite stable and the only changes that have been required have been just incorporation of the latest improvements to icon.library etc. as they occur. Some of the proposals in this thread, eg. MCP, MUI5, etc. are much more likely to cause issues than BB3&4 is.

[utri007]

I haven't tried BB3/4 for a long time. Previous problems migh well be solved, like buggy libraries and devices. Like graphics.library and/or scsi.device etc.

Thing is that it ruins also a startup-sequence and I don't want that.

[Locutus]

I have run Executive since...forever, works well and especially with high cpu load keeps the system smoother.

Also it gives CLI process tools (ps, top, renice) which i like. Never had compatibility issues.

[Minuous]

@utri007:

Yes, those components were removed (in the case of graphics.library) or replaced (in the case of scsi.device) years ago.

Re. startup-sequence, it lets the user decide: either to update it automatically, or it can just bring up EditPad and tell you the necessary changes to make. Either way the startup-sequence is not "ruined"!

[kolla]

Originally Posted by PeterK

Are you sure? In my opinion, Executive is only recommended for multiuser systems, like the good old BBS in the 1990s. I would expect that it brings no advantages on a typical single user system. It might just cause more overhead for the multitasking.

Executive is a good tool for people who know what they are doing, who know what they want and why. Yeah, multiuser environment is one use case where it is handy, but not the only use case. It's great for anyone who runs lots of background tasks that require CPU, such as rendering, graphics and audio converting/decoding/encoding etc.

Hi @Minuous,

Ok, cool. I'll give them a shot in that case. Is there a list somewhere of what all is included/patched/changed with a default bb3+4 installation?

Thanks!

Also @Minuous, I notice if I use a default A4000 setup in FS-UAE (EC030 cpu) after installing BB3+4 and selecting my computer model as an A4000D it will no longer boot, it remains in a black screen. However if I change my CPU to a full 040 it does boot.

Ok, quick status update, have installed and am having good results so far with the following:
- PFS3
- OS3.9 BB1-4
- MMULIB with MuMove4k, MuFastZero, MuProtectModules
- SaferPatches
- PoolMem
- TrueMultiAssigns
- Env-Handler

A bit more performance tuning and I think I should be at a decent baseline to begin with usability/look and feel tweaks.

I'm thinking of calling this project "Zen Workbench", or "ZenWB", or "zwb" for short. Thoughts?

Ideally if I can make this all work the way that I want I'd like to make it reusable, in a workbench pack. That's a stretch goal for now but one that I think could be cool.

[Minuous]

@esc:

>Is there a list somewhere of what all is included/patched/changed with a default bb3+4 installation?

http://amigan.1emu.net/releases/BoingBags3&4.readme

>I notice if I use a default A4000 setup in FS-UAE (EC030 cpu) after installing BB3+4 and selecting my computer model as an A4000D it will no longer boot, it remains in a black screen. However if I change my CPU to a full 040 it does boot.

Thanks, I will investigate the cause. I assume you are using V1.36?

Yep!

[kolla]

MULib may not be a good choice as not all systems have MMU, for example the EC030 systems.

[daxb]

Originally Posted by kolla

MULib may not be a good choice as not all systems have MMU, for example the EC030 systems.

From guide:

[...]The 68EC030 is a cheaper variant of
systems. the 68030 without a MMU, but the mmu.library will be smart enough to detect this chip and will fail on a 68EC030. [...] Using the mmu.library on systems without a MMU is possible, but doesn't make much sense,[...]

[Minuous]

@esc:

OK, the cause is the updated scsi.device (http://aminet.net/package/driver/media/SCSI4345p) which is used if you choose A4000D and specify that you want SCSI.device updated. Without an MMU it will not work, but this is not mentioned in the scsi patch's readme.

BTW generally emulator uses should choose "Other" as the model as they don't need scsi.device updated. But this is only mentioned if you press the Help button at that point in the installer.

Either doing a reinstall and choosing "Other" as the model, or just copying over Files2/S/startup-sequence.none or Files4/S/startup-sequence.none to s:startup-sequence (depending on whether you installed the contributions (4) or not (2)), will fix the problem.

I have submitted a formal bug report.

[Toni Wilen]

Jumping to conclusions too quickly?.. How did you narrow it down to that update? (I'd expect loading non-patched scsi.device using same method to have exact same behavior)

Also if you choose 68030 in UAE, you won't actually get real EC030 model (until next version) so this test is meaningless without real hardware (and I don't know what FS-UAE does with it). And finally, "real" EC030 does not really exist because there are EC030's that have partially functional MMU and can't be software detected.

[Minuous]

>Jumping to conclusions too quickly?

Most likely, yes, sorry :-(

>How did you narrow it down to that update?

I assumed that the unpatched one would work on all real A4000s, or emulations of A4000s, otherwise they would not be able to boot their systems to install BB3&4 in the first place.

>I'd expect loading non-patched scsi.device using same method to have exact same behavior

Do you mean there something wrong with the LoadModule call? It should not require an MMU AFAIK.