I wanted to send a note about a problem I encountered when compiling
vectorial.c from Cs4 CVS because I am not sure that I resolved it correctly.
This is more of a C issue than a Csound one I suppose, but I am in slightly
unfamiliar territory here.

Basically, the following code would not compile for me with CodeWarrior
because of the pointer arithmetic in the last two statements below.

typedef struct {
        OPDS    h;
        MYFLT   *ifnOut, *ifnIn, *ifnDel, *ielements, *imaxd, *istod;
        AUXCH   aux;
        MYFLT   **buf, *outvec, *invec, *dlyvec;
        long    *left, maxd;
        int     elements;
} VECDEL;

void vecdly_set(VECDEL *p)
{
    FUNC        *ftp;
    int elements = (p->elements = (int) *p->ielements), j;
[....]

      if (p->aux.auxp == NULL ||
          (int)(elements * sizeof(MYFLT *)
                + n * elements * sizeof(MYFLT)
                + elements * sizeof(long)) > p->aux.size) {
        auxalloc(elements * sizeof(MYFLT *)
                 + n * elements * sizeof(MYFLT)
                 + elements * sizeof(long),
                 &p->aux);
        p->buf= (MYFLT **) p->aux.auxp;
        for (j = 0; j < elements; j++) {
          p->buf[j] = (MYFLT *) (p->aux.auxp + sizeof(MYFLT *)* elements
                                 +sizeof(MYFLT ) * n * j);
        }
        p->left= (long *)  (p->aux.auxp +sizeof(MYFLT *)* elements
                            +sizeof(MYFLT ) * n * elements);
      }


p->aux.auxp is of type void* and the compiler (rightly, I think) complains
that adding an integer to a void* is illegal (since it is impossible to know
the size of what it points to).  I was able to get it to compile by adding
casts to (MYFLT **), like this:

p->buf[j] = (MYFLT *) ((MYFLT **)p->aux.auxp + sizeof(MYFLT *)* elements
                                 +sizeof(MYFLT ) * n * j);

p->left= (long *)  ((MYFLT **)p->aux.auxp +sizeof(MYFLT *)* elements
                            +sizeof(MYFLT ) * n * elements);

But the more that I think about it, the more that I think (MYFLT **) is
wrong and it should probably be (char*).  (Adding one to a MYFLT** pointer
moves 4 bytes -- the size of a pointer -- and adding one to a char* pointer
should move one byte, right?)

So my question is, what is the correct way to fix this problem so that the
code works correctly??

Thanks very much for the help.  And sorry to be asking a language issue
here.

Anthony Kozar
anthonykozar@sbcglobal.net
http://akozar.spymac.net/



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Anthony Kozar wrote:

> But the more that I think about it, the more that I think (MYFLT **) is
> wrong and it should probably be (char*).  (Adding one to a MYFLT** pointer
> moves 4 bytes -- the size of a pointer -- and adding one to a char* pointer
> should move one byte, right?)
> 
> So my question is, what is the correct way to fix this problem so that the
> code works correctly??

If a compiler allows adding to a void* pointer, then it usually
assumes that sizeof(void)==1, so the cast to char* should work.
However, the code is wrong for another reason as well: the resulting
MYFLT* and long* pointers may not be correctly aligned on 64 bit
machines with sizeof(MYFLT)==4, and n*j or n*elements being odd:
     p->buf[j] = (MYFLT *) (p->aux.auxp + sizeof(MYFLT *)* elements
                            +sizeof(MYFLT ) * n * j);
     p->left= (long *)  (p->aux.auxp +sizeof(MYFLT *)* elements
                         +sizeof(MYFLT ) * n * elements);
By the way, in old versions of Csound (on which CsoundAV is based),
AUXCH.auxp used to be char* and not void*, and this may be the source
of the problem, but as mentioned above there are also alignment issues
on 64 bit platforms (on which CsoundAV was obviously not intended to
be run).


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Thanks very much for the explanation, Istvan!

I will go with the (char*) cast for now and think about the alignment issue
for the future.  (Perhaps the Cs5 version of vectorial.c can deal with this,
but I am not sure that worrying about it for Cs4 will matter ...).

I imagine the alignment issue could be solved with something like this
though (done before calling auxalloc):

if  ( n%2 )  ++n;

Anthony Kozar
anthonykozar@sbcglobal.net
http://akozar.spymac.net/


On 4/10/05 6:09 PM, Istvan Varga  etched in stone:

> the resulting
> MYFLT* and long* pointers may not be correctly aligned on 64 bit
> machines with sizeof(MYFLT)==4, and n*j or n*elements being odd:



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

 From the looks of it, casting to char* is probably what you want, as
you are adding bytes to p->aux.auxp, instead of items. I suppose
if the latter was the case, the original would read like this:

  p->buf[j] = (MYFLT *) (p->aux.auxp +  elements
                                  + n * j);

In fact, in CS.h, we see that

typedef struct auxch {
         struct auxch * nxtchp;
         long   size;
         void   *auxp, *endp;    /* was char* */
} AUXCH;

at one point *auxp was a pointer to char*.

I suppose vectorial.c is using an old definition of AUXCH

Victor

At 22:36 10/04/2005, you wrote:
>I wanted to send a note about a problem I encountered when compiling
>vectorial.c from Cs4 CVS because I am not sure that I resolved it correctly.
>This is more of a C issue than a Csound one I suppose, but I am in slightly
>unfamiliar territory here.
>
>Basically, the following code would not compile for me with CodeWarrior
>because of the pointer arithmetic in the last two statements below.
>
>typedef struct {
>         OPDS    h;
>         MYFLT   *ifnOut, *ifnIn, *ifnDel, *ielements, *imaxd, *istod;
>         AUXCH   aux;
>         MYFLT   **buf, *outvec, *invec, *dlyvec;
>         long    *left, maxd;
>         int     elements;
>} VECDEL;
>
>void vecdly_set(VECDEL *p)
>{
>     FUNC        *ftp;
>     int elements = (p->elements = (int) *p->ielements), j;
>[....]
>
>       if (p->aux.auxp == NULL ||
>           (int)(elements * sizeof(MYFLT *)
>                 + n * elements * sizeof(MYFLT)
>                 + elements * sizeof(long)) > p->aux.size) {
>         auxalloc(elements * sizeof(MYFLT *)
>                  + n * elements * sizeof(MYFLT)
>                  + elements * sizeof(long),
>                  &p->aux);
>         p->buf= (MYFLT **) p->aux.auxp;
>         for (j = 0; j < elements; j++) {
>           p->buf[j] = (MYFLT *) (p->aux.auxp + sizeof(MYFLT *)* elements
>                                  +sizeof(MYFLT ) * n * j);
>         }
>         p->left= (long *)  (p->aux.auxp +sizeof(MYFLT *)* elements
>                             +sizeof(MYFLT ) * n * elements);
>       }
>
>
>p->aux.auxp is of type void* and the compiler (rightly, I think) complains
>that adding an integer to a void* is illegal (since it is impossible to know
>the size of what it points to).  I was able to get it to compile by adding
>casts to (MYFLT **), like this:
>
>p->buf[j] = (MYFLT *) ((MYFLT **)p->aux.auxp + sizeof(MYFLT *)* elements
>                                  +sizeof(MYFLT ) * n * j);
>
>p->left= (long *)  ((MYFLT **)p->aux.auxp +sizeof(MYFLT *)* elements
>                             +sizeof(MYFLT ) * n * elements);
>
>But the more that I think about it, the more that I think (MYFLT **) is
>wrong and it should probably be (char*).  (Adding one to a MYFLT** pointer
>moves 4 bytes -- the size of a pointer -- and adding one to a char* pointer
>should move one byte, right?)
>
>So my question is, what is the correct way to fix this problem so that the
>code works correctly??
>
>Thanks very much for the help.  And sorry to be asking a language issue
>here.
>
>Anthony Kozar
>anthonykozar@sbcglobal.net
>http://akozar.spymac.net/
>
>
>
>-------------------------------------------------------
>SF email is sponsored by - The IT Product Guide
>Read honest & candid reviews on hundreds of IT Products from real users.
>Discover which products truly live up to the hype. Start reading now.
>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
>_______________________________________________
>Csound-devel mailing list
>Csound-devel@lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/csound-devel

Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Could anyone update the linux snapshot at the sourceforge CVS (jpff?)?
Thanks a lot,


Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I had been neglecting that.  Do you just want the binary or the tar as well?
==John ffitch


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Just the tarball sources. Thanks a lot!

Victor

At 10:43 11/04/2005, you wrote:
>I had been neglecting that.  Do you just want the binary or the tar as well?
>==John ffitch
>
>
>-------------------------------------------------------
>SF email is sponsored by - The IT Product Guide
>Read honest & candid reviews on hundreds of IT Products from real users.
>Discover which products truly live up to the hype. Start reading now.
>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
>_______________________________________________
>Csound-devel mailing list
>Csound-devel@lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/csound-devel

Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

p->buf[j] = (MYFLT *) ((MYFLT **)p->aux.auxp + sizeof(MYFLT *)* elements
                                 +sizeof(MYFLT ) * n * j);

Surely that should be

p->buf[j] = (MYFLT *)p->aux.auxp + elements + n*j;

or is the data structure over odd?  Have a FOO** and not following an
indirection and then casting to a FOO* is extremely contorted.
==John ffitch


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Victor Lazzarini wrote:

> Just the tarball sources. Thanks a lot!

But if you only need the sources, you can always get that from CVS.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I can't do CVS at the moment.

Victor

At 12:35 11/04/2005, you wrote:
>Victor Lazzarini wrote:
>
>>Just the tarball sources. Thanks a lot!
>
>But if you only need the sources, you can always get that from CVS.
>
>
>-------------------------------------------------------
>SF email is sponsored by - The IT Product Guide
>Read honest & candid reviews on hundreds of IT Products from real users.
>Discover which products truly live up to the hype. Start reading now.
>http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
>_______________________________________________
>Csound-devel mailing list
>Csound-devel@lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/csound-devel

Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I think it's the way it is because it is trying to partition space from
auxalloc() into several different arrays, not all of the same type.  I
originally added the cast (MYFLT **) because of this line in the code:

        p->buf= (MYFLT **) p->aux.auxp;

Then later I began to think that this was wrong.  Sorry for any confusion :)

Anthony

On 4/11/05 7:00 AM, jpff@codemist.co.uk  etched in
stone:

> p->buf[j] = (MYFLT *) ((MYFLT **)p->aux.auxp + sizeof(MYFLT *)* elements
> +sizeof(MYFLT ) * n * j);
> 
> Surely that should be
> 
> p->buf[j] = (MYFLT *)p->aux.auxp + elements + n*j;
> 
> or is the data structure over odd?  Have a FOO** and not following an
> indirection and then casting to a FOO* is extremely contorted.
> ==John ffitch



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net