"implementation of `malloc'" because I have not the time to
test/check the original malloc.c code in glibc.
Linux uses `mremap' to reallocate large blocks, perhaps the answer is
there but I'm not sure without a test/reading. I have followed my
instinct for the first solution and the reason for the second.

I have written

> Note: `tmp = instrument+1' causes segfault on arch-32bit, so the offset is >=2.

therefore your third solution in git

  instr = (INSTR*)realloc(instr, (instrument+1)*sizeof(INSTR));

fails (segfault) on arch-32bit with

; by Brian Baughn 3-14-05
; bbaughn@berklee.net

beats = 100
permeasure = 4

i101    m1 b1   q    mp   D3
i101                      F3
i101                      D3

i101    m2 b1             D3
i101       b3             D3

i101    m3 b1             D3
i101                      F3
i101                      D3

i101    m4 b1             D3
i101       b3             D3

i101    m5 b1              D3
i101       b4              G5

i101    m6 b1              E5
i101       b2              F5
i101       b3     e      Eb5
i101       b3.5   e          
i101       b4     q          

i101    m7 b1     e       D5
i101                            
i101                q          
i101                e      Db5
i101                            
i101                q          

i101    m8 b1     q       D5
i101                         E5
i101                         D5

i102    m1 b2      q      D4
i102       b4              E4
i102       b4             Bb3

i102    m2 b2              F4
i102       b2              B3
i102       b4             C#4
i102       b4             Bb3

i102    m3 b2      q      D4
i102       b4              E4
i102       b4             Bb3

i102    m4 b2              F4
i102       b2              B3
i102       b4             C#4
i102       b4             Bb3

i103    m5 b2      e     F6
i103       b2      e     A5
i103       b2.5    e     D6
i103       b3      e     F6
i103       b3      e     A5
i103       b4      e     E6

i103    m6 b1      q     C#6
i103               q     D6
i103               e     C6
i103                           
i103               q        

i103    m7 b1      e     B5
i103                            
i103                 q         
i103                 e    Bb5
i103                            
i103                 q         

i103     m8 b1     e     F5
i103        b1     e     A5
i103        b1.5   e     D6
i103        b2     e    Bb5
i103        b2.5   e     D6
i103        b3     q     F5
i103        b3           A5

i104     m8 b1     e     F5
end

tito

On Tue, May 29, 2012 at 03:34:47PM +0100, jpff@cs.bath.ac.uk wrote:
> Can I have a less cryptic explanation.  As far as I can see the change was
> to swap two fields in a struct.  I cannot see how that changes anything
> other than (invisible) layout.
> 
> If you are saying that it is a fenceposts -- count from zero -- I can
> blieve it is possible
> ==John
> 
> >> Can anyone explain why this changes anything?
> >
> > implementation of `malloc'
> >
> > PS  the second solution seems better

------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net