ChangeFSI broken

ChangeFSI is broken with the latest ROM 02-Jun-14 on the pandaboard.
When you try and load a jpeg in ChangeFSI you get “Bad Register”

The previous ROM 29-May-14 is OK, looks like this is to do with the recent BASIC changes :-(

ChangeFSI is broken with the latest ROM 02-Jun-14 on the pandaboard.
When you try and load a jpeg in ChangeFSI you get “Bad Register”

I get the same error message relating to a chunk of inline assembler in the RunImage of !Blackhole.
Specifically it’s the iconbar sprite change routine “PROCassemble”

The previous ROM 29-May-14 is OK, looks like this is to do with the recent BASIC changes :-(

Snap, I’m thinking the same thing.

Where did you get Blackhole from, I’ve found v2 but it’s not 32bit compat because of the module :-(
I assume you’ve got v1?

When you try and load a jpeg in ChangeFSI you get “Bad Register”

Looks like I didn’t think through the optimisation of “a number greater than 9 is implicitly greater than 5”. In the spirit of optimisation my fix for that is to delete another instruction! Try again with tomorrow’s ROM.

Where did you get Blackhole from, I’ve found v2 but it’s not 32bit compat because of the module :-(
I assume you’ve got v1?

No, Blackhole 2, just a few modifications for the beagle (which doesn’t have ADFS and complains about that, but the core function works after a small hack) and the Hole module needs a small alteration to work with RO5.
I’ve been running it as a test on RPCEmu and the beagle since October.

The hole350 module just needs a tidy replacement for the MOVS PC,LR and a 32 bit header. When I did it I went for a crude MOVS —> MOV. While that works, it probably isn’t the neatest answer so I think I will run it through again . It keeps me amused when there’s nothing decent on the TV¹ and I want to forget about work.

¹ Which is most of the time

Having a good recycle bin is one of the things I miss most about RO4 and RO6 (and, I feel, is one of the biggest omissions of RO5). For many years I used Blackhole under RO4 – would you be able share your modified version? If so, would it run on a Raspberry Pi? (I’d be very happy to test it out :-).

Part of the copyright bit of Blackhole2 states: “If you wish to use pieces of my code in your own pd programs then please feel free to do so, provided these programs are not intented (sic) for sale.” So perhaps there wouldn’t be too much of a problem there, providing the original author was credited?

The hole350 module just needs a tidy replacement for the MOVS PC,LR and a 32 bit header.

What version of BlackHole is that? My copy has a Hole500 module and works on the Iyonix.

So perhaps there wouldn’t be too much of a problem there, providing the original author was credited?

Alas, just above the bit that you quoted the conditions on distribution state: “b) That the application is supplied unchanged.”

What version of BlackHole is that? My copy has a Hole500 module and works on the Iyonix.

Blackhole 2 version 4.18

Never seen a later version. It doesn’t have a Hole500. What’s your source?

Alas, just above the bit that you quoted the conditions on distribution state: “b) That the application is supplied unchanged.”

Yes, there is quite some ambivalence in the legal bits but I prefer to go with the spirit of the thing :-)

Seriously, it’s such a good application maybe it’s worth finding out what the score is direct from the author, Mark Greenwood, if he’s still contactable. That is of course if Steve P is OK about it all?

There’s a lot of nice little apps like this (memphis being one), is BSD the preferred licence?
I’m not saying we should, but if we can have licences so that it is possible if wanted to put some more apps in the default disc image we could

Never seen a later version. It doesn’t have a Hole500. What’s your source?

It was on one of the last Foundation RISC User CDs. Annoyingly it’s also marked 4.18, the Hole500 module is dated 18 Apr 2004

Yes, there is quite some ambivalence in the legal bits but I prefer to go with the spirit of the thing :-)

The spirit is that you must distribute the application unchanged. The fact that you are allowed to use bits of the software in your own projects is completely separate from that.

Contacting authors to try and keep software alive is always a good idea as we’ve lost way too much good stuff already. It may even yield an unexpected result, I recently contacted Alisdair McDiarmid about Transient’s licence and ended up with the copyright being transferred to me.

The spirit is that you must distribute the application unchanged.

Did you miss my smiley? Perhaps it was a joke that software authors would find in poor taste – in which case I apologise – but, nevertheless, it was a joke.

Anyway, I’ve emailed Mark Greenwood at the address given at the end of BlackHole’s Legal & Copyright bit. If the email address is still used by him and if he responds I’ll let you know what the outcome is.

Fixed, thanks.

Seriously, it’s such a good application maybe it’s worth finding out what the score is direct from the author, Mark Greenwood, if he’s still contactable. That is of course if Steve P is OK about it all?

Tried emailing last year. No response.
BTW. Not my copyright so nothing to do with me except that under euro law I’m entitled to modify the program to fix an obvious fault, but only for my use. I’m not entitled to distribute.

If you think about it a combination of the Hole filer delete interception and Freds “Transient” would be a better bet.

Ideally the filer delete being intercepted would produce a captured file with path in the recycle bin so that a restore to original location is easier. Which of course means neither Blackhole, nor I think Transient, fit the bill exactly.

Fixed, thanks.

Confirmed. Nice quick fix.

If you think about it a combination of the Hole filer delete interception and Freds “Transient” would be a better bet.
Ideally the filer delete being intercepted would produce a captured file with path in the recycle bin so that a restore to original location is easier. Which of course means neither Blackhole, nor I think Transient, fit the bill exactly.

The best Recycle bin I’ve found has been !Recycler from Fabis (and now APDL?). It allows restore to original location, allows for multiple deletions of the same object, traps overwrites as well as deletes and has a whitelist of locations not to trap from. If it could be made RO5 compatible, it would be ideal, in my view.

Alternatively, if the OS could provide the relevant hooks to allow easy trapping of deletes and overwrites I guess a new one could be made.

Alternatively, if the OS could provide the relevant hooks to allow easy trapping of deletes and overwrites I guess a new one could be made.

The Hole module in Blackhole already does both deletes and overwrites.
Trap File Overwrites is the second tick box down in the Singularity options menu – slightly confusing I suppose as the menu says Singularity and the module is called Hole

Did you miss my smiley? Perhaps it was a joke that software authors would find in poor taste – in which case I apologise – but, nevertheless, it was a joke.

No, I didn’t miss the smiley but your earlier quote along with going with the spirit rather than what the licence actually said, seemed to suggest that you thought taking liberties was all right. Apologies if I read too much into it.

If you think about it a combination of the Hole filer delete interception and Freds “Transient” would be a better bet.

That’s actually one of the additions to Transient that Alisdair was thinking of, but he never finished it. Trapping Filer deletes is easy, the hard part is finding out which top level objects are about to be deleted.

Sitting on FileV or upcall 3,6 (which is what BlackHole does) gives you a bottom up view of deletions. Just enable ‘Trap Filer Deletes’ in BlackHole then delete an application to see what that results in.

What you really want is to move an application or directory in its entirety to the bin, keeping its structure intact. I guess the easiest way would be to exrend FilerAction, as per ROL branch, it probably could be done without that I think but with some difficulty.

The best Recycle bin I’ve found has been !Recycler from Fabis

How did it handle applications/directories? Were they kept intact or did you end up with a bunch of individual files?

Sitting on FileV or upcall 3,6 (which is what BlackHole does) gives you a bottom up view of deletions. Just enable ‘Trap Filer Deletes’ in BlackHole then delete an application to see what that results in.

Yeah, hence my comment about needing to capture the path info.
Without the structure the result in the bin is akin to a collection of boiled sweets released from the jar – flat…

keeping its structure intact. I guess the easiest way would be to exrend FilerAction, as per ROL branch, it probably could be done without that I think but with some difficulty.

Certainly – otherwise how can you tell the difference between “user is deleting these files within !App” and “user is deleting !App”?

How did it handle applications/directories? Were they kept intact or did you end up with a bunch of individual files?

I wonder if this isn’t something that a “bin” could handle at its UI end? While the files might appear as a disparate list of files, if the containing directory itself should become one of the deleted objects, the bin software could consolidate everything within into one item, to permit you to undelete each file individually, or the whole lot in one click.

A good test, perhaps, is to create a directory “test”, copy in half a dozen files (001, 002 … 006). Delete the directory.
Then create the directory “test”, copy in half a dozen different files (001, 002 … 006). Delete that.
Then look and see what is in the bin. :-)
A useful bin will attempt to hold “generations”.
A really useful one will trap overwrites and copy out the original file before it is overwritten (allowing a “revert” feature).

How did it handle applications/directories? Were they kept intact or did you end up with a bunch of individual files?

It just creates a pile of files, not files in a directory structure, sadly.

Last time I tried Blackhole (a long time ago, mind), I didn’t find the ‘trap overwrites’ feature, so moved on. Does it have a ‘whitelist’ too, so you can tell it not to trap from certain locations (eg. !Scrap)?

Yeah, hence my comment about needing to capture the path info.

Yes, if you keep track of the full paths then you can restore the original app/dir and if you implement Rick’s suggestion of adding some cleverness to the UI you wouldn’t have to hand-pick all the constituent files.

However if objects could be dragged to the bin too you’d end up with individual files if you used Filer delete but with an intact app/dir when dragging to the bin – unless you’d split drags up in separate files too (which seems silly to me). The same goes for OS_File 6 vs OS_FSControl 27.

Personally I’d prefer the bin to deal with top level objects as much as possible, it makes for a tidier bin and possibly a simpler UI. Nevertheless I’ll have a think about Rick’s suggestions.

to permit you to undelete each file individually, or the whole lot in one click.

How would it handle restoring a file that’s nested inside a deleted app/dir? Recreate the higher level directories or say the file can’t be restored.

It just creates a pile of files, not files in a directory structure, sadly.

Thanks, so it’s basically the same as BlackHole in this regard.

Does it have a ‘whitelist’ too, so you can tell it not to trap from certain locations (eg. !Scrap)?

Yes, you can tell it to not trap certain paths.

Sounds like BlackHole is pretty much what’s wanted, apart from the keeping directory structures intact then. The only problem seems to be that there’s no way of getting hold of a fully RPi compatible version now, if I follow the thread correctly?

One possible solution is to distribute the application unchanged along with another application that makes the necessary changes to the binary in the first application.