AmigaOS 3.2 and beyond

[Olaf Barthel]

Originally Posted by Daedalus

I'm pretty sure all Amigas start counting at 1.1.1978. Usually the system time is taken from the boot volume if no RTC is present though - possibly the filesystem creation date (date of formatting)?

As usual, not only is this more complicated than most people (including myself) might expect, it's also a very obscure feature to boot.

The default file system (which is the FFS in Kickstart 2.04 and beyond, and the original BCPL file system in older Kickstart 1.x versions) may set the system time according to what it finds on the boot volume. The boot volume can be a floppy disk drive, hard disk or even RAD: for that matter.

The system time may be changed if the current time seems to be "unset". Which means that the number of days since January 1st 1978 is 0 and the number of minutes since midnight is 0, too. In that case, the default file system will look into the root directory of the boot volume and take look at one of three date stamps which can be found there.

The three date stamps are: 1) when was the directory last altered, 2) when was the disk last altered and 3) when was the disk formatted. The default file system uses the second of the three ("disk altered") as the new system time, if it needs to and uses the timer.device/TR_SETSYSTIME command for this purpose.

You might wonder what the "disk altered" date stamp represents and what will actually result in updating it. This date stamp only gets updated if the disk's bitmap is changed. This means that merely modifying a file (overwrite it) or directory (rename it, change its comment, etc.) will not update it. Deleting or creating a file or directory will, and so will appending data to a file, provided that at least as many bytes are written as are needed to allocate another block.

To wrap this up: this feature is available only in the old Amiga file system or the FFS (and the OS4 FFS reimplementation by yours truly). No third party file systems are required to support it.

[Predseda]

[Madowax]

Sadly after ROM 3.2 installation internal 880k df1 on my A4000 does not work anymore. It was working ok on 3.1 and 3.1.4

[Steady]

That seems strange. I would first check that you didn't disconnect anything on your df1: drive when you were installing the new ROM. Maybe your cable is flipped or the power is loose?

Does it even get seen? To check this, hold down both mouse keys on boot. Then select "Boot Options" and see if df1 is in the list.

I am suspecting a hardware issue rather than a software one initially.

[Madowax]

Originally Posted by Steady

That seems strange. I would first check that you didn't disconnect anything on your df1: drive when you were installing the new ROM. Maybe your cable is flipped or the power is loose?

Does it even get seen? To check this, hold down both mouse keys on boot. Then select "Boot Options" and see if df1 is in the list.

I am suspecting a hardware issue rather than a software one initially.

So did I and I double checked every connections before posting, I have even tried a couple of different drives. Nevertheless I will make some more testing. It simply gives me the df1 error like when tlhere is no ready signal from the drive. Same goes for kirf floppy test utility. Both drives works if configured as df0:
Edit:
I will replace the cable at this point, probably it is gone for good. Clearly it should happen when you replace the ROMs

[Swe_Kryten2x4b]

Originally Posted by boemann

I like the tab completion very much myself. Scrollback is not a 3.2 feature, but it will make for a nice addition in 3.2.1 or later, so Jason and I are working on it as we speak. There is absolutely no promises and it it is far from a small task so we will see if it happens at all. But at least we are giving it a try.

Sounds great! I really miss that part from KingCON so I hope it happens. I realize I could just install KingCON but since the new shell works fine using tab completion otherwise I may or may not do it.

Another suggestion. Unless I've missed it, it would be nice to have a qualifier key for scrolling sideways using a scroll-wheel mouse.

[boemann]

Originally Posted by Swe_Kryten2x4b

Another suggestion. Unless I've missed it, it would be nice to have a qualifier key for scrolling sideways using a scroll-wheel mouse.

Ah I did think about it but forgot about it again - well nothing a commodity can't be made to do - I'll leave than as an exercise for some developer out there.

[Thomas Richter]

Originally Posted by Swe_Kryten2x4b

Sounds great! I really miss that part from KingCON so I hope it happens.

You can get that already today. Use ViNCEd, and the boot shim, and you even have it as boot console. Yes, it does leave TAB expansion to the shell, and no longer does that through the console (which was quite an abuse of design).



Actually, the v47 TAB expansion mechanism is the ViNCEd mechanism all along, which is a much better separation of shell and console responsibilities than what we had before.

[Pyromania]

It looks like my order of AmigaOS 3.2 has just shipped from Amiga On The Lake. I'm excited to get it and do a fresh install.

[kamelito]

IIRC Olsen did update at least MagiMenu so why is it not include with AmigaOS 3.2?
A nice update to it should be to be able to move to any item in the menu using the wheel mouse that way no need to move the mouse to select an entry.

[Swe_Kryten2x4b]

Originally Posted by Thomas Richter

You can get that already today. Use ViNCEd, and the boot shim, and you even have it as boot console. Yes, it does leave TAB expansion to the shell, and no longer does that through the console (which was quite an abuse of design).



Actually, the v47 TAB expansion mechanism is the ViNCEd mechanism all along, which is a much better separation of shell and console responsibilities than what we had before.

You know, I've tried ViNCEd in the past but there was something about it that bugged me, that I've now of course forgotten what it was. I'm going to give it another go though! Now installed again and will use it thoroughly and see if it still bugs me...

That boot shim is a great idea. Unfortunately, my computer doesn't like it.Sure, I do get the ViNCEd shell in boot with no startup-sequence but when I boot normally it crashes, and I never can boot into workbench.

I suspect why though. It doesn't like HappyENV.

I like to keep ram: clean so only the things I explicitly put there is there, which is why I'm not a huge fan of either the old copy everything from envarc: to env: in ram, or the new softlink approach.

You can't get everything I suppose...

[Romanujan]

Is there any chance to see updated ("reactionized") Prefs in the next release? The ones from OS 3.9 look so much better...

[daxb]

Originally Posted by Swe_Kryten2x4b

I suspect why though. It doesn't like HappyENV.

Try envhandler instead. http://aminet.net/package/util/sys/envhandler

[Thomas Richter]

Originally Posted by Swe_Kryten2x4b

That boot shim is a great idea. Unfortunately, my computer doesn't like it.Sure, I do get the ViNCEd shell in boot with no startup-sequence but when I boot normally it crashes, and I never can boot into workbench.

I suspect why though. It doesn't like HappyENV.

Please don't suspect... please provide a bug report. The boot menu has a nice option to trace through the startup-sequence. Activate this please, and let me know at which point you see a crash.


BTW, if you use the boot shim, please *remove* the "SetVNC Mount Override as CON:" line, if you have one in user-startup. That's no longer required and counter-productive.


Thanks,
Thomas

[Swe_Kryten2x4b]

Originally Posted by daxb

Try envhandler instead. http://aminet.net/package/util/sys/envhandler

That was my first attempt, since that was I was using before in 3.5, 3.9, and 3.1.4.

But just get "can't create env:" or words to that effect. Now I tried again, and now it worked. Go figure.

With envhandler it works using ViNCEds con-handler! Sometimes a recoverable alert on boot. I will need to use it like this for a while to see how frequent that is.

[Swe_Kryten2x4b]

Originally Posted by Thomas Richter

Please don't suspect... please provide a bug report. The boot menu has a nice option to trace through the startup-sequence. Activate this please, and let me know at which point you see a crash.


BTW, if you use the boot shim, please *remove* the "SetVNC Mount Override as CON:" line, if you have one in user-startup. That's no longer required and counter-productive.


Thanks,
Thomas

Will do. It works better wiith envhandler though (but not pefect), but I will do the same with that enabled.

There's nothing in user-startup regarding con: override.

Btw, of course I want it to run perfectly on my system but it's unreasonable for me (or anyone for that matter) to expect you to be able to take every little patch people like to use into account. Especially for people like me who always uses heavily modified s-s.

[bubbob42]

Originally Posted by Romanujan

Is there any chance to see updated ("reactionized") Prefs in the next release? The ones from OS 3.9 look so much better...

Well, you picked one of the three Prefs we didn’t touch - Font. Coincidence?

Compare the others. I never liked the old 3.9 prefs, some look like as if one had strewn gadgets across an empty window, but tastes differ, of course. That’s why I almost replicated ASL and WorkbenchPrefs for 3.1.4 to make them look as identical as possible compared to their 3.9 counterparts.

The 3.2 WorkbenchPrefs now look like a crossbreed of 3.9 and 4.1, due to added functionality and written using GadTools. Not an ideal solution, but fast.

There will be Reaction-based prefEditors if someone writes them. Some Prefs (like IControl) will have to be rewritten, because their windows won’t fit NTSC 640x200 anymore (which they have to) and they already have several subwindows.

But honestly, how many times will you change prefs? There‘s a lot of more important things to improve.

[Romanujan]

Originally Posted by bubbob42

Well, you picked one of the three Prefs we didn’t touch - Font. Coincidence?

Compare the others.(...)

Coincidence. They typically share the same problems:
- they all lack the window close gadget, which is irritating
- they are limited to fixed-width fonts, and they are using the same one as the console window; I don't know any single font that would look nice both within console (or ASCII editor) and within normal application window - the OS 3.9 prefs fixed this
- many of them lack any graphics, or the graphics is very simplified (Input, Locale)
- many of them are simply too large; for example, the font selection one is almost as large as OS 3.9 one, but (contrary to 3.9 one) does not show the preview; another example is ASL, which is twice as big as 3.9 one. IControl, Input, Palette (BTW - this one has a preview, but it does not consider window bar / screen title bar font/size settings), Reaction, Serial, DefaultIcons - these are IMHO way too big
- they use old-style gadgets

I'm reinstalling the OS (my previous installation was very messy, it was my old 3.9 installation migrated to 3.1.4), and I'm trying different settings (fonts, etc.) to "refresh" everything, so I'm spending quite a lot of time with various preferences. This is definitely not as pleasant as it was with OS 3.9.

[Warty]

Amiga on the Lake says my CD-ROM is coming on Monday, so excited about that! I downloaded the NDK just now, and am starting to look through it, but of course without an actual OS to compile for, it's a very theoretical exercise. But one question I did have: if we convert our apps to Reaction, they won't work on systems without Reaction. As I understand it, Hyperion actually bought the rights to Reaction. Any chance that Reaction can be made available to non-3.2 systems, so that our apps work on older systems?

Maybe it's dependent on something in the KS, or another updated library, which would make it a no go from a technical point of view.

[Minuous]

Originally Posted by Warty

Any chance that Reaction can be made available to non-3.2 systems, so that our apps work on older systems?

ReAction has been part of the OS for over 20 years now, since OS3.5.