Bug OS3.2.1 clipboard with sas/c se editor

[hceline]

Posting here because hyperion forum will not let me log in.


There seems to be an incompatibility with the clipboard handling in OS 3.2.1 compared to 3.1.


Under 3.2.1 in sas/c editor se I can copy from shell to se with the "Block->Input from Clip" menu.

The other way using "Block->Output to Clip" does not work when pasting into con or other programs. It gets copied to the clipboard, "Block->Input from Clip" pastes the text back into se, but pasting into any other programs silently fails.

Only exception is BlackEdtior from aminet, who report an "Clipboard I/O-error" on Paste. (But BlackEditor is apparently not to be trusted as it would happily crash entire WinUAE under 3.1)



Under 3.1, coping both ways works.
OS version between 3.1.and 3.2.1 not tested.

[freakofnature]

I would like to help, but can't. Every part of SAS/C (editor, options, etc) crashes on startup when running from WB 3.2. I stopped trying.
For SAS/C i'll stick to WB 3.9.

[hceline]

Update: The info in this post is not to be trusted. There was something wrong with setup I used. After setting up a fresh install of 3.2.1 other muForce hits and various errors disappeared. And I have not been able to reproduce this hit on the new install so far.


Did some testing with MuForce, MuGuardianAngel and PatchWork.
First thing I noticed is that if se is started directly from tool-icon and not from a project-icon it will throw Enforcer hit on both 3.1 and 3.2.1 when editing/marking text.
No matter how it is started it will throw one PatchWork hit on exec.library FindPort() called without Forbid() during initialization under both 3.1 and 3.2.1.
But on 3.2.1 only, when starting from a copy of sas/examples/hello.c and selection the printf line with mouse from behind for copying, I get this hit during selection process.

*MuForce Hit
BYTE READ from 046F7D1F                        PC: 04695E14
USP : 046ABA20 SR: 0000  (U0)(-)(-)  TCB: 046A9F48
Data: 00010006 00000074 00000001 00000000 00008003 00000001 00000002 00000000
Addr: 046E7D18 0464BED0 04647514 00000000 046EF318 00000000 04647560 040022A4
Stck: 00000000 00000002 00000000 04695AF4 FFFFFFFF FFFFFFFF 00000000 00000000
Stck: 00000000 00000000 011AAFB7 00000000 00000000 04647534 046475E4 04647560
----> 04695E14 - "sc:c/se"  Hunk 0000 Offset 00000BC4
----> 04695AF4 - "sc:c/se"  Hunk 0000 Offset 000008A4
04695df2 :  6100 1f8c                  bsr $4697d80
04695df6 :  7800                       moveq.l #$0,d4
04695df8 :  3800                       move.w d0,d4
04695dfa :  2004                       move.l d4,d0
04695dfc :  5280                       addq.l #$1,d0
04695dfe :  6604                       bne.s $4695e04
04695e00 :  6100 0c0a                  bsr $4696a0c
04695e04 :  0c84 0000 0100             cmpi.l #$100,d4
04695e0a :  6d1a                       blt.s $4695e26
04695e0c :  2004                       move.l d4,d0
04695e0e :  d080                       add.l d0,d0
04695e10 :  41ec 8a00                  lea.l -$7600(a4),a0
04695e14 : *0830 0003 0801             btst #$3,$1(a0,d0.l)
04695e1a :  670a                       beq.s $4695e26
04695e1c :  206c 731e                  movea.l $731e(a4),a0
04695e20 :  6100 1564                  bsr $4697386
04695e24 :  7a00                       moveq.l #$0,d5
04695e26 :  6100 3258                  bsr $4699080
04695e2a :  4a80                       tst.l d0
04695e2c :  6704                       beq.s $4695e32
04695e2e :  6100 0bdc                  bsr $4696a0c
04695e32 :  7200                       moveq.l #$0,d1
Name: "sc:c/se"  Hunk 0000 Offset 00000BC4

[Minuous]

SAS/C works here under OS3.2(.1), including the editor, compiler and linker. The SCOPTS program doesn't work but it did not work on OS3.9 either.

I have investigated the clipboard bug. SE is writing a malformed IFF file to the clipboard; specifically, the length of the CHRS chunk is incorrect. Some programs will ignore this malformation but most will not.

[hceline]

Thanks for investigating.
Under 3.1 the malformed IFF just works it seems, I could paste it anywhere I tried.

Regarding the Enforcer hit I previous post, a little more testing indicates Picasso96 is the culprit; not OS 3.2.1. Have not tested OS 3.1 with Picasso96, but the error goes away when I boot OS 3.2.1 to native modes.

On the other hand, except the Clipboard and Picasso96 issues, I can report that the whole sas/c 6.58 package works under 3.2.1 here. This includes the debugger and SCOPTS, with the only caveat that debugger must be started from Shell (Will do lots of byte reads from $00000000 if started from Workbench) and editor must be started from a project-icon.

[a/b]

Clipboard bug patched (v6.58), se.lha is in the zone.
Details... there are 3 lengths: I/O, IFF chunk, raw text data. The last two are calculated incorrectly, with extra rounding complications for the last one.

[freakofnature]

I created four setups for testing, KS 3.1(WB 3.9) and KS 3.2(WB 3.2) in both FS-UAE and WinUAE.
Under FS-UAE SAS/C crashes all the time running KS 3.2. KS 3.1 runs fine with PIV, crashes with UAEgfx.
Using WinUAE no crashes with 3.2, all works nicely. I realized i got two different hardfiles with KS 3.1. One of it makes SAS/C crashing immediately.

As hceline said, there is something wrong with Picasso96 in one of its versions.

Thanks a/b, will test later.

[hceline]

Originally Posted by a/b

Clipboard bug patched (v6.58), se.lha is in the zone.
Details... there are 3 lengths: I/O, IFF chunk, raw text data. The last two are calculated incorrectly, with extra rounding complications for the last one.

Output to Clip works under OS 3.2.1 now, thank you.

Update: The info below is not to be trusted. There was something wrong with setup I used. After setting up a fresh install of 3.2.1 other muForce hits and various errors disappeared. And I have not been able to reproduce this hit on the new install so far.

Edit2: Everything below might be irrelevant. The Enforcer hit described here seems be be provoked by PatchWork. When running only MuForce And MuGuardianAngel it does not happen. But the Enforcer hit described earlier when selecting text with mouse does happen without PatchWork and only under OS 3.2.1 and Picasso96, not under 3.1 and Picasso96.

But for the Enforcer hits under Picasso96 the plot thickens. The editor works on under OS 3.1 in 32bit BRGA, 16bit PC and 8bit. Under OS 3.2.1 in any of those modes, if I open a file and edit it with keyboard and then just move the mousepointer into the window, I get this form MuForce and PatchWorks:

*MuForce Hit
LONG READ from 1202000A                        PC: 041B8794
USP : 043313B8 SR: 0018  (U0)(F)(-)  TCB: 0432FA48
Data: 041BB7FF 000000FF 00000001 0000FFFF 00008000 00000000 043313D0 C00000FF
Addr: 041BB682 12020000 041BB6C4 00000000 041BB6B2 04301714 04000884 040022A4
Stck: 04344838 00000000 FFFFE66C 0000000A 000000E6 000001B4 00000000 041BB74D
Stck: 00000000 00000000 00000001 0000FFFF 00008000 00000001 80000000 C0000000
----> 041B8794 - "PatchWork"  Hunk 0000 Offset 0000078C
----> 04344838 - "sc:c/se"  Hunk 0000 Offset 000086F8
----> 041BB74D - "PatchWork"  Hunk 0000 Offset 00003745
041b8774 :  4241                       clr.w d1
041b8776 :  4841                       swap.l d1
041b8778 :  0800 0003                  btst #$3,d0
041b877c :  6706                       beq.s $41b8784
041b877e :  c2bc 0000 00ff             and.l #$ff,d1
041b8784 :  0800 0004                  btst #$4,d0
041b8788 :  6702                       beq.s $41b878c
041b878a :  e589                       lsl.l #$2,d1
041b878c :  2f01                       move.l d1,-(a7)
041b878e :  60b8                       bra.s $41b8748
041b8790 :  226d 0038                  movea.l $38(a5),a1
041b8794 : *2f29 000a                  move.l $a(a1),-(a7)
041b8798 :  224f                       movea.l a7,a1
041b879a :  4eb9 041b c464             jsr $41bc464
041b87a0 :  2e46                       movea.l d6,a7
041b87a2 :  2f05                       move.l d5,-(a7)
041b87a4 :  41fa 0124                  lea.l $41b88ca(pc),a0
041b87a8 :  224f                       movea.l a7,a1
041b87aa :  4eb9 041b c464             jsr $41bc464
041b87b0 :  588f                       addq.l #$4,a7
041b87b2 :  4cdf 0102                  movem.l (a7)+,d1/a0
Name: "sc:c/se"

 SetPointer(0x04344838,0x00000000,-6548,10,230,436)
Severity 0: hot spot is outside of the sprite
PC=18B4042B TCB=0432FA48 ("sc:c/se")
Data: 042FE66C 0000000A 02D500E6 00CF01B4 00F60052 04000300 2002320F 04381A80
Addr: 04344838 00000000 00000000 0000042F 78C80430 C828040B 12020000 00000430
Stck: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000

If I type some more and then move the mouse, the excact same errors repeats.

Edit: This is with Picasso96 version 3.3.2 (29.08.2022)