QLiberator decompiler

[mk79]

If I run the compiled Diskmate 5 in QPC2, it stops with an error on Q_ERR_ON, which I assume is this line. You can then 'C'ontinue to get past it. So I thought it was an issue with Q_ERR_ON in one of the QLiberator extensions.

The FEX that DiskMate expects is a procedure. The FEX that is now part of SMSQ/E is a function. Q_ERR_ON only works with procedures... as tofro said, name clash.

[tofro]

From the FileInfo 2 Manual:

The FEX procedure is the exact equivalent of FI2_CALL and FEX_W is equivalent to
FI2_CALL with the bit 31 of the key set. FEX and FEX_W are kept for upward
compatibility with older FileInfo things only; these procedures should not be
used any more in modern programs as their name clashes with SMSQ/E v3+ built-in
PROC/FN. By default, FileInfo II v3.50+ is configured so to skip the
initialization of the FEX and FEX_W procedures.

So - Simply replace "FEX" with "FI2_CALL" in the DiskMate source code and you should be set, and DiskMate should be working with newer SMSQ/E versions.

Tobias

[BSJR]

From the FileInfo 2 Manual:

The FEX procedure is the exact equivalent of FI2_CALL and FEX_W is equivalent to
FI2_CALL with the bit 31 of the key set. FEX and FEX_W are kept for upward
compatibility with older FileInfo things only; these procedures should not be
used any more in modern programs as their name clashes with SMSQ/E v3+ built-in
PROC/FN. By default, FileInfo II v3.50+ is configured so to skip the
initialization of the FEX and FEX_W procedures.

So - Simply replace "FEX" with "FI2_CALL" in the DiskMate source code and you should be set, and DiskMate should be working with newer SMSQ/E versions.

Tobias

I doubt this will work. DM5 expects the name FEX and will not react to FI2_CALL.
For that DM5 will need to be edited and recompiled.
I have changed the name FEX to FFX in both my copies of FI2 and DM5 and configured FI2 to use both names.
Now DM5 will not complain.

BSJR

[tofro]

F
So - Simply replace "FEX" with "FI2_CALL" in the DiskMate source code and you should be set, and DiskMate should be working with newer SMSQ/E versions.

Tobias

I doubt this will work. DM5 expects the name FEX and will not react to FI2_CALL.
For that DM5 will need to be edited and recompiled.
I have changed the name FEX to FFX in both my copies of FI2 and DM5 and configured FI2 to use both names.
Now DM5 will not complain.

BSJR

Of course it will - maybe there's a slight misunderstanding. Replacing that single occurrence is obviously not enough, rather all occurrences of FEX in DM5 need to be replaced. I thought that was clear. Then this will end up to be the same solution as yours, just without changing FI2 as well.

Tobias

[Martin_Head]

Yesterday I tried to get the decompiled QLiberator to compile itself. I wondered, that if I got the right compile options, I could maybe make an exact copy the the original QLIB_OBJ. That way I would know that the decompile was accurate.

Didn't go quite to plan. It would not compile itself. It got so far, then BASIC threw up an error message. So I am now bug hunting for either a decompiler problem, or if I have made a typo somewhere while tidying up the program code.

[pjw]

Im sure you dont need any advice from me Even so, since I use this program quite a lot, I venture some suggestions in case you hadnt got round to them yet: Try compiling with the already compiled QLib. For debugging purposes switch on Names, Lines, and Stats (The final version will probably have all these switched off). Try compiling from Qdos rather than SMSQ/E. DP's Xref program can be useful for finding bugs and inconsistencies. The Turbo parser can also be useful when all else fails (ie try compiling it with Turbo just for its analysis)..

[EmmBee]

once.zip

(25.61 KiB) Downloaded 97 times

I've attached my "once" program. What it can do is to identify any names used once only - which could be misspellings. It has to be run in QPC2 as it uses the functions DEVTYPE, SCR_XLIM and SCR_YLIM, which are only available in SMS. I hope this helps. Best of luck.

[Martin_Head]

A bit of an update.

The BASIC error was caused by a typo on my part, rather than a problem with the decompile. That got the compile to work.
The next problem, was that when I EXEC the resulting object I got a Qliberator error, something about an undefined variable. That turned out to be CMD$. I had to identify the variable in the source code, and rename it CMD$.
Next, When I EXECed the the program (The one that compiled itself, and one compiled with the original QLiberator) I could not select any of the input boxes, or buttons. This turned out to be because I had the WINDS selected during compile, which seems to stop the pointer interface bits working.

At the moment, I think the recompiled QLIB is maybe 10 bytes shorter that the original, The dataspace maybe 32 bytes bigger. If I use QLIB_PATCH, it reports the number of channels as 10, when it should be 9. When you use a compiler directive of $$chan= Should it be 1 less than the number you want, to allow for channel 0?

There is one other odd thing I have noticed when comparing the original QLIB with my recompiled one's. There is a bit of data stored between the end of name list, and the start of the name/variable table

This is the start of the program, note the something odd???

Code: Select all

00282A70  60000082       bra       $00282AF4                   `...
00282A74  00004AFB       ori.b     #$FB,d0                     ..J.
00282A78  0009           ILLEGAL   INSTRUCTION                 ..
00282A7A  516C6962       subq.w    #$08,$6962(a4)              Qlib
00282A7E  5F332E33       subq.b    #$07,$33(a3,d2.l)           _3.3
00282A82  3620                                                 6 
00282A84  2020                           
00282A86  2020                                
00282A88  2020                              
00282A8A  2020                                
00282A8C  2020                               
00282A8E  2020                                
00282A90  00000146       start of SuperBASIC program  282BB6
00282A94  0000B18C       name/variable table          28DBFC
00282A98  00000000       line number table - none     
00282A9C  00040000       
00282AA0  736E           
00282AA2  0009           channels  9
          0080           buffer? 128
          02000000   
00282AAA  1B58           heap size 7000
          0394       
00282AAE  0330
00282AB0  0000AECA       name list                    28D93A
00282AB4  00009630       start of extensions          28C0A0
00282AB8  00004FA2       dataspace?  20386
00282ABC  0000157C       stack?     5500
00282AC0  0001
          0000B16C       something odd???   28DBDC
          0001       
00282AC8  FF00           
00282ACA  00000000       ori.b     #$00,d0                     ....
00282ACE  00000000       ori.b     #$00,d0                     ....

And this is the something odd. Any one recognize what it may be, and how to incorporate into the compile

Code: Select all

0028DBD4  065F4845       addi.w    #$4845,(a7)+                ._HE
0028DBD8  4C50           ILLEGAL   INSTRUCTION                 LP
0028DBDA  4600           not.b     d0                          F.
end of name list


something odd
0028DBDC  00000000       ori.b     #$00,d0                     ....
0028DBE0  00020010       ori.b     #$10,d2                     ....
0028DBE4  04 514C494200  "QLIB"
0028DBEA  00000000       ori.b     #$00,d0                     ....
0028DBEE  00004E71       ori.b     #$71,d0                     ..Nq
0028DBF2  283C0000000C   move.l    #$0000000C,d4               (<....
0028DBF8  6000FD36       bra       $0028D930                   `..6


name/variable table
0028DBFC  0318           btst      d1,(a0)+                    ..
0028DBFE  00001800       ori.b     #$00,d0                     ....
0028DC02  0618000B       addi.b    #$0B,(a0)+                  ....
0028DC06  1800           move.b    d0,d4                       ..
0028DC08  1118           move.b    (a0)+,-(a0)                 ..
0028DC0A  00181800       ori.b     #$00,(a0)+                  ....
0028DC0E  1F18           move.b    (a0)+,-(a7)                 ..
0028DC10  00231800       ori.b     #$00,-(a3)                  .#..

[EmmBee]

The manual says: "The minimum size of a channel table is 3 entries, for channels 0, 1 and 2" - so its the actual number of channels that need to be given.
CMD$ is initially set up by QLiberator as a variable to read the incoming Option command string and can be used as a normal variable afterwards.

Sounds like you're making excellent progress. Are you compiling with QDOS or SMSQ/E?

[pjw]

This is the start of the program, note the something odd???

I had a bit of a fiddle, but no joy. Sorry.
Fantastic progress, though!