ABC - Problem with Q: field

[Roger Hare]

I am using EasyABC on a HP machine running Windoze 10.

 

I have a problem with correct representation of numeric values in the Q: field. 

 

Briefly, any value above 127 produces an incorrect value both in the display, and in the generated PDF file.
Here's an example:

 

X:1
T:Newcastle
M:4/4
L:1/8
Q:1/4=150
K:G
|: A2 | "G" B2 d2 "D7" G2 A2 | "G" G3 A G2 D2 | "G" B2 d2 G2 d2 | "C" e2 g4 fe | 
 "G" d2 B2 A2 G2 | "C" E2 e4 dc | "G" d2 B2 "D7" A3 G |1 "G" G4 G2 :|2 "G" G6 |] 
|: ef | "G" gfed g3 B | "Am" A2 g4 A2 | "Em" G3 A "Bm" B2 F2 | "Em" E2 e4 f2 | 
 "G" gfed g3 B | "Am" A2 A2 c3 d | "C" e2 B2 "D7" A3 G | "G" G6 :|

 

I've attached the PDF generated by this ABC code - the metronome marking which should be 150 is
shown as -106.

 

A value of 127 works fine; a value of 128, yields -128; a value of 129 yields -127; and so on... (*)

I'm baffled by this - particularly as on my other W10 machine, it all works fine. 

 

Any ideas?

 

Thank you. (I thought I had already asked this question, but I can't find any trace...)

 

Roger

(*) It hasn't escaped my attention that 127=(2**7)-1, and that the value produced is a discontinuous function of the input Q: value:
        f(x)_1:127=x; 
        f(x)_128:256=x-256
       which just shows that I'm a right smarty-knickers, but which doesn't get me much forrader...  ?

 

Newcastle.pdf

Edited April 5, 2019 by lachenal74693

[alex_holden]

1 hour ago, lachenal74693 said:

(*) It hasn't escaped my attention that 127=(2**7)-1, and that the value produced is a discontinuous function of the input Q: value:
        f(x)_1:127=x; 
        f(x)_128:256=x-256
       which just shows that I'm a right smarty-knickers, but which doesn't get me much forrader...  ?

 

 

You're on the right track. Whatever program EasyABC is using to render the ABC (abc2ps?) is storing the Q value in an unsigned 8 bit variable, without correctly validating that the number will fit in a variable of that size.

[alex_holden]

Looking at this header file, tempo is defined as a short, which is usually 16 bits, not 8. Perhaps the compiler that was used for the DOS version of abcm2ps is treating short as 8 bits? Or maybe it's an older version that is using char instead of short? The parser function is reading the value into an int, then blindly copying it into a short (not good practice). I would be inclined to just change the type in the header file to an int.

[alex_holden]

I just tried abcm2ps on MacOS (installed via Macports) and it correctly processes Q: values higher than 127, which suggests an issue with the DOS version included with EasyABC for Windows. Just for kicks, I tried giving it a tempo of 32768, and as expected it printed -32768 in the output (showing that it's storing the number in a signed 16 bit short variable).

[John Wild]

I have EasyABC version 1.3.5, running on Windows 10 home.

I pasted the ABC text above into a file and exported a PDF without changing anything.

The score displayed correctly with Q=150.    Newcastle.pdf

I can offer no explanation of why your program produced a different result.

You could try entering Q:150instead of Q:1/4=150. It may not make a difference but anything is worth trying.

[David Barnert]

5 hours ago, lachenal74693 said:

(I thought I had already asked this question, but I can't find any trace...)

 

I think so, too. It sounds familiar, from a long time ago.

 

You don’t really dance Newcastle at 150, do you?

[Roger Hare]

10 hours ago, David Barnert said:

I think so, too. It sounds familiar, from a long time ago.

 

You don’t really dance Newcastle at 150, do you?

 

Heh, heh, no - I just set it to 150 as an example which was >128...

 

Problem solved - I have version 1.3.7.7 installed, but the problem was with abcm2ps in version 1.3.7.6

(https://sourceforge.net/p/easyabc/tickets/12/), and I had foolishly got the 1.3.7.7 file settings pointing to

the old 1.3.6.6 copies of abcm2ps, etc. Correcting the file settings fixed the problem. What a dolt ? - I'm

normally a bit more careful than that!

 

Thank you for the responses folks.

 

Roger

Edited April 6, 2019 by lachenal74693