As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
cigoto? Anyway, it should be a simple fix.

Dr Victor Lazzarini
Senior Lecturer
Dept. of Music
NUI Maynooth Ireland
tel.: +353 1 708 3545
Victor dot Lazzarini AT nuim dot ie




------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I think if it's emitting a kngoto, then the relevant i-time opcode
would be cngoto. One thought I had was that maybe the semantic
analyzer should check the rates of opcodes within the branch, and
really determine which to use depending on if any of the opcodes are
run at perf-time or not.  That seems like it would solve a lot of
Csound coding issues.  Thoughts?

On Wed, Aug 7, 2013 at 1:15 PM, Victor Lazzarini
 wrote:
> As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
> a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
> cigoto? Anyway, it should be a simple fix.
>
> Dr Victor Lazzarini
> Senior Lecturer
> Dept. of Music
> NUI Maynooth Ireland
> tel.: +353 1 708 3545
> Victor dot Lazzarini AT nuim dot ie
>
>
>
>
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

The relevant code seems to be in csound_orc_expressions.c:

TREE * create_goto_token(CSOUND *csound, char * booleanVar,
                         TREE * gotoNode, int type)
{
/*     TREE *ans = create_empty_token(csound); */
    char* op = (char *)csound->Malloc(csound, 7); /* Unchecked */
    TREE *opTree, *bVar;

    switch(gotoNode->type) {
    case KGOTO_TOKEN:
      sprintf(op, "ckgoto");
      break;
    case IGOTO_TOKEN:
      sprintf(op, "cigoto");
      break;
    case ITHEN_TOKEN:
      sprintf(op, "cngoto");
      break;
    case THEN_TOKEN:
    case KTHEN_TOKEN:
      sprintf(op, "cngoto");
      break;
    default:
      if (type) sprintf(op, "ckgoto");
      else sprintf(op, "cggoto");
    }


you see that there is no difference between ITHEN, THEN or KTHEN. 
The oentry for cngoto is:

 { "cngoto", S(CGOTO),0,   3,      "",     "Bl",   ingoto, kngoto, NULL     },

with 3 as thread, which is OK for THEN. KTHEN, but not ITHEN.

Should this be cigoto for ITHEN?


On 7 Aug 2013, at 12:48, Steven Yi wrote:

> I think if it's emitting a kngoto, then the relevant i-time opcode
> would be cngoto. One thought I had was that maybe the semantic
> analyzer should check the rates of opcodes within the branch, and
> really determine which to use depending on if any of the opcodes are
> run at perf-time or not.  That seems like it would solve a lot of
> Csound coding issues.  Thoughts?
> 
> On Wed, Aug 7, 2013 at 1:15 PM, Victor Lazzarini
>  wrote:
>> As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
>> a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
>> cigoto? Anyway, it should be a simple fix.
>> 
>> Dr Victor Lazzarini
>> Senior Lecturer
>> Dept. of Music
>> NUI Maynooth Ireland
>> tel.: +353 1 708 3545
>> Victor dot Lazzarini AT nuim dot ie
>> 
>> 
>> 
>> 
>> ------------------------------------------------------------------------------
>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>> It's a free troubleshooting tool designed for production.
>> Get down to code-level detail for bottlenecks, with <2% overhead.
>> Download for free and get started troubleshooting in minutes.
>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
> 
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead. 
> Download for free and get started troubleshooting in minutes. 
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

Dr Victor Lazzarini
Senior Lecturer
Dept. of Music
NUI Maynooth Ireland
tel.: +353 1 708 3545
Victor dot Lazzarini AT nuim dot ie




------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I just took a look.  cigoto won't do it as it's looking for the
positive case, while cngoto check for the negative case.  We'd need to
modify or add an opcode that runs just ingoto and no kngoto. I think
ultimately we should do as mentioned and have the semantic analyzer
choose the correct option depending on if the opcodes within the
branch are i-time only or not.  That will be some work though, due to
nested if's.

On Wed, Aug 7, 2013 at 2:40 PM, Victor Lazzarini
 wrote:
> The relevant code seems to be in csound_orc_expressions.c:
>
> TREE * create_goto_token(CSOUND *csound, char * booleanVar,
>                          TREE * gotoNode, int type)
> {
> /*     TREE *ans = create_empty_token(csound); */
>     char* op = (char *)csound->Malloc(csound, 7); /* Unchecked */
>     TREE *opTree, *bVar;
>
>     switch(gotoNode->type) {
>     case KGOTO_TOKEN:
>       sprintf(op, "ckgoto");
>       break;
>     case IGOTO_TOKEN:
>       sprintf(op, "cigoto");
>       break;
>     case ITHEN_TOKEN:
>       sprintf(op, "cngoto");
>       break;
>     case THEN_TOKEN:
>     case KTHEN_TOKEN:
>       sprintf(op, "cngoto");
>       break;
>     default:
>       if (type) sprintf(op, "ckgoto");
>       else sprintf(op, "cggoto");
>     }
>
>
> you see that there is no difference between ITHEN, THEN or KTHEN.
> The oentry for cngoto is:
>
>  { "cngoto", S(CGOTO),0,   3,      "",     "Bl",   ingoto, kngoto, NULL     },
>
> with 3 as thread, which is OK for THEN. KTHEN, but not ITHEN.
>
> Should this be cigoto for ITHEN?
>
>
> On 7 Aug 2013, at 12:48, Steven Yi wrote:
>
>> I think if it's emitting a kngoto, then the relevant i-time opcode
>> would be cngoto. One thought I had was that maybe the semantic
>> analyzer should check the rates of opcodes within the branch, and
>> really determine which to use depending on if any of the opcodes are
>> run at perf-time or not.  That seems like it would solve a lot of
>> Csound coding issues.  Thoughts?
>>
>> On Wed, Aug 7, 2013 at 1:15 PM, Victor Lazzarini
>>  wrote:
>>> As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
>>> a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
>>> cigoto? Anyway, it should be a simple fix.
>>>
>>> Dr Victor Lazzarini
>>> Senior Lecturer
>>> Dept. of Music
>>> NUI Maynooth Ireland
>>> tel.: +353 1 708 3545
>>> Victor dot Lazzarini AT nuim dot ie
>>>
>>>
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>> It's a free troubleshooting tool designed for production.
>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>> Download for free and get started troubleshooting in minutes.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> Csound-devel mailing list
>>> Csound-devel@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>
>> ------------------------------------------------------------------------------
>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>> It's a free troubleshooting tool designed for production.
>> Get down to code-level detail for bottlenecks, with <2% overhead.
>> Download for free and get started troubleshooting in minutes.
>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>
> Dr Victor Lazzarini
> Senior Lecturer
> Dept. of Music
> NUI Maynooth Ireland
> tel.: +353 1 708 3545
> Victor dot Lazzarini AT nuim dot ie
>
>
>
>
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

OK, for the moment, I followed your suggestion and added a new oentry (cingoto) and
changed csound_orc_expressions.c to use it.
in GIT now.
On 7 Aug 2013, at 23:14, Steven Yi wrote:

> I just took a look.  cigoto won't do it as it's looking for the
> positive case, while cngoto check for the negative case.  We'd need to
> modify or add an opcode that runs just ingoto and no kngoto. I think
> ultimately we should do as mentioned and have the semantic analyzer
> choose the correct option depending on if the opcodes within the
> branch are i-time only or not.  That will be some work though, due to
> nested if's.
> 
> On Wed, Aug 7, 2013 at 2:40 PM, Victor Lazzarini
>  wrote:
>> The relevant code seems to be in csound_orc_expressions.c:
>> 
>> TREE * create_goto_token(CSOUND *csound, char * booleanVar,
>>                         TREE * gotoNode, int type)
>> {
>> /*     TREE *ans = create_empty_token(csound); */
>>    char* op = (char *)csound->Malloc(csound, 7); /* Unchecked */
>>    TREE *opTree, *bVar;
>> 
>>    switch(gotoNode->type) {
>>    case KGOTO_TOKEN:
>>      sprintf(op, "ckgoto");
>>      break;
>>    case IGOTO_TOKEN:
>>      sprintf(op, "cigoto");
>>      break;
>>    case ITHEN_TOKEN:
>>      sprintf(op, "cngoto");
>>      break;
>>    case THEN_TOKEN:
>>    case KTHEN_TOKEN:
>>      sprintf(op, "cngoto");
>>      break;
>>    default:
>>      if (type) sprintf(op, "ckgoto");
>>      else sprintf(op, "cggoto");
>>    }
>> 
>> 
>> you see that there is no difference between ITHEN, THEN or KTHEN.
>> The oentry for cngoto is:
>> 
>> { "cngoto", S(CGOTO),0,   3,      "",     "Bl",   ingoto, kngoto, NULL     },
>> 
>> with 3 as thread, which is OK for THEN. KTHEN, but not ITHEN.
>> 
>> Should this be cigoto for ITHEN?
>> 
>> 
>> On 7 Aug 2013, at 12:48, Steven Yi wrote:
>> 
>>> I think if it's emitting a kngoto, then the relevant i-time opcode
>>> would be cngoto. One thought I had was that maybe the semantic
>>> analyzer should check the rates of opcodes within the branch, and
>>> really determine which to use depending on if any of the opcodes are
>>> run at perf-time or not.  That seems like it would solve a lot of
>>> Csound coding issues.  Thoughts?
>>> 
>>> On Wed, Aug 7, 2013 at 1:15 PM, Victor Lazzarini
>>>  wrote:
>>>> As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
>>>> a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
>>>> cigoto? Anyway, it should be a simple fix.
>>>> 
>>>> Dr Victor Lazzarini
>>>> Senior Lecturer
>>>> Dept. of Music
>>>> NUI Maynooth Ireland
>>>> tel.: +353 1 708 3545
>>>> Victor dot Lazzarini AT nuim dot ie
>>>> 
>>>> 
>>>> 
>>>> 
>>>> ------------------------------------------------------------------------------
>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>> It's a free troubleshooting tool designed for production.
>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>> Download for free and get started troubleshooting in minutes.
>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>> _______________________________________________
>>>> Csound-devel mailing list
>>>> Csound-devel@lists.sourceforge.net
>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>> 
>>> ------------------------------------------------------------------------------
>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>> It's a free troubleshooting tool designed for production.
>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>> Download for free and get started troubleshooting in minutes.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> Csound-devel mailing list
>>> Csound-devel@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>> 
>> Dr Victor Lazzarini
>> Senior Lecturer
>> Dept. of Music
>> NUI Maynooth Ireland
>> tel.: +353 1 708 3545
>> Victor dot Lazzarini AT nuim dot ie
>> 
>> 
>> 
>> 
>> ------------------------------------------------------------------------------
>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>> It's a free troubleshooting tool designed for production.
>> Get down to code-level detail for bottlenecks, with <2% overhead.
>> Download for free and get started troubleshooting in minutes.
>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
> 
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead. 
> Download for free and get started troubleshooting in minutes. 
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

Dr Victor Lazzarini
Senior Lecturer
Dept. of Music
NUI Maynooth Ireland
tel.: +353 1 708 3545
Victor dot Lazzarini AT nuim dot ie




------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I'm not sure I understand 100% the implications of this, but are you
thinking of changing how the goto operates based on what is inside the
relevant (goto'ed) section? I would think it possible (intentionally)
that one create an igoto block that may have k-rate opcodes in it.
Sorry for the noise if I misunderstood the intention of the change.
best
Oeyvind

2013/8/8 Steven Yi :
> I just took a look.  cigoto won't do it as it's looking for the
> positive case, while cngoto check for the negative case.  We'd need to
> modify or add an opcode that runs just ingoto and no kngoto. I think
> ultimately we should do as mentioned and have the semantic analyzer
> choose the correct option depending on if the opcodes within the
> branch are i-time only or not.  That will be some work though, due to
> nested if's.
>
> On Wed, Aug 7, 2013 at 2:40 PM, Victor Lazzarini
>  wrote:
>> The relevant code seems to be in csound_orc_expressions.c:
>>
>> TREE * create_goto_token(CSOUND *csound, char * booleanVar,
>>                          TREE * gotoNode, int type)
>> {
>> /*     TREE *ans = create_empty_token(csound); */
>>     char* op = (char *)csound->Malloc(csound, 7); /* Unchecked */
>>     TREE *opTree, *bVar;
>>
>>     switch(gotoNode->type) {
>>     case KGOTO_TOKEN:
>>       sprintf(op, "ckgoto");
>>       break;
>>     case IGOTO_TOKEN:
>>       sprintf(op, "cigoto");
>>       break;
>>     case ITHEN_TOKEN:
>>       sprintf(op, "cngoto");
>>       break;
>>     case THEN_TOKEN:
>>     case KTHEN_TOKEN:
>>       sprintf(op, "cngoto");
>>       break;
>>     default:
>>       if (type) sprintf(op, "ckgoto");
>>       else sprintf(op, "cggoto");
>>     }
>>
>>
>> you see that there is no difference between ITHEN, THEN or KTHEN.
>> The oentry for cngoto is:
>>
>>  { "cngoto", S(CGOTO),0,   3,      "",     "Bl",   ingoto, kngoto, NULL     },
>>
>> with 3 as thread, which is OK for THEN. KTHEN, but not ITHEN.
>>
>> Should this be cigoto for ITHEN?
>>
>>
>> On 7 Aug 2013, at 12:48, Steven Yi wrote:
>>
>>> I think if it's emitting a kngoto, then the relevant i-time opcode
>>> would be cngoto. One thought I had was that maybe the semantic
>>> analyzer should check the rates of opcodes within the branch, and
>>> really determine which to use depending on if any of the opcodes are
>>> run at perf-time or not.  That seems like it would solve a lot of
>>> Csound coding issues.  Thoughts?
>>>
>>> On Wed, Aug 7, 2013 at 1:15 PM, Victor Lazzarini
>>>  wrote:
>>>> As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
>>>> a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
>>>> cigoto? Anyway, it should be a simple fix.
>>>>
>>>> Dr Victor Lazzarini
>>>> Senior Lecturer
>>>> Dept. of Music
>>>> NUI Maynooth Ireland
>>>> tel.: +353 1 708 3545
>>>> Victor dot Lazzarini AT nuim dot ie
>>>>
>>>>
>>>>
>>>>
>>>> ------------------------------------------------------------------------------
>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>> It's a free troubleshooting tool designed for production.
>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>> Download for free and get started troubleshooting in minutes.
>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>> _______________________________________________
>>>> Csound-devel mailing list
>>>> Csound-devel@lists.sourceforge.net
>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>
>>> ------------------------------------------------------------------------------
>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>> It's a free troubleshooting tool designed for production.
>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>> Download for free and get started troubleshooting in minutes.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> Csound-devel mailing list
>>> Csound-devel@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>
>> Dr Victor Lazzarini
>> Senior Lecturer
>> Dept. of Music
>> NUI Maynooth Ireland
>> tel.: +353 1 708 3545
>> Victor dot Lazzarini AT nuim dot ie
>>
>>
>>
>>
>> ------------------------------------------------------------------------------
>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>> It's a free troubleshooting tool designed for production.
>> Get down to code-level detail for bottlenecks, with <2% overhead.
>> Download for free and get started troubleshooting in minutes.
>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel



-- 

Oeyvind Brandtsegg
Professor of Music Technology
NTNU
7491 Trondheim
Norway
Cell: +47 92 203 205

http://flyndresang.no/
http://www.partikkelaudio.com/
http://soundcloud.com/brandtsegg
http://soundcloud.com/t-emp

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Hi Oeyvind,

Nope, you got it exactly right.  The intention is to change depending
on what is inside the block. My rationale behind it is as follows:

If's work by jumping over blocks of opcodes.  However, with the case
of something like:


if (ival > 0) ithen
    krateval someOpcode ival, 0
else
    krateval someOpcode ival, 1
endif

what happens is that you get something like this:

#b0 lt ival, 0
cingoto #b0, __synth_label0
krateval someOpcode ival, 0
igoto __synth_label1
:__synth_label0
krateval someOpcode ival, 1
:__synth_label1

Now, if the ival happens to be greater than zero, you'll get what
seems like the right behavior. During init-time, the (ival > 0) is
evaluated and stored into #b0.  cingoto checks and decides not to
jump, then the first someOpcode is initialized.  Then the goto will
jump past the rest of the opcodes to synth_label1, and the 2nd
someOpcode is not initialized.  At perf-time, everything *seems* to be
fine.  What happens is cingoto has no perf function, so it will pass
through to the first branch *always*.  The first branch is run and the
rest is jumped over with the goto.  However, all seems fine.

Now, what happens if ival is <= 0?  The cingoto will evaluate and
because it is not true, jump to synth_label0.  The 2nd someOpcode now
is initialized, and the first is not.

Now at perf time, we're in trouble.  Because cingoto has no perf-time
function, evaluation falls through and we will always go through the
first branch of code.  The first branch has not be initialized, and we
get bad behavior. In the case of the attached test, we get a crash
with a SEGFAULT.


Now, my analysis is this: if you have an i-time boolean evaluation
with an if-then, when you have opcodes that perform at k-rate, it can
and will cause errors.  Up until now, we seem to have gotten around
those errors by just always running at k-rate.  So if you wrote code
with i-time boolean and k-rate opcodes, that was not going to cause
issues.  However, with i-time only evaluation, we have the
possibilities for serious problems, as illustrated above and in the
example CSD.

Note, opcodes that take in k-var's and output k-var's may not
necessarily be k-rate opcodes.  For example, i(kval) is one, and
chnexport.

steven


On Thu, Aug 8, 2013 at 8:47 AM, Oeyvind Brandtsegg
 wrote:
> I'm not sure I understand 100% the implications of this, but are you
> thinking of changing how the goto operates based on what is inside the
> relevant (goto'ed) section? I would think it possible (intentionally)
> that one create an igoto block that may have k-rate opcodes in it.
> Sorry for the noise if I misunderstood the intention of the change.
> best
> Oeyvind
>
> 2013/8/8 Steven Yi :
>> I just took a look.  cigoto won't do it as it's looking for the
>> positive case, while cngoto check for the negative case.  We'd need to
>> modify or add an opcode that runs just ingoto and no kngoto. I think
>> ultimately we should do as mentioned and have the semantic analyzer
>> choose the correct option depending on if the opcodes within the
>> branch are i-time only or not.  That will be some work though, due to
>> nested if's.
>>
>> On Wed, Aug 7, 2013 at 2:40 PM, Victor Lazzarini
>>  wrote:
>>> The relevant code seems to be in csound_orc_expressions.c:
>>>
>>> TREE * create_goto_token(CSOUND *csound, char * booleanVar,
>>>                          TREE * gotoNode, int type)
>>> {
>>> /*     TREE *ans = create_empty_token(csound); */
>>>     char* op = (char *)csound->Malloc(csound, 7); /* Unchecked */
>>>     TREE *opTree, *bVar;
>>>
>>>     switch(gotoNode->type) {
>>>     case KGOTO_TOKEN:
>>>       sprintf(op, "ckgoto");
>>>       break;
>>>     case IGOTO_TOKEN:
>>>       sprintf(op, "cigoto");
>>>       break;
>>>     case ITHEN_TOKEN:
>>>       sprintf(op, "cngoto");
>>>       break;
>>>     case THEN_TOKEN:
>>>     case KTHEN_TOKEN:
>>>       sprintf(op, "cngoto");
>>>       break;
>>>     default:
>>>       if (type) sprintf(op, "ckgoto");
>>>       else sprintf(op, "cggoto");
>>>     }
>>>
>>>
>>> you see that there is no difference between ITHEN, THEN or KTHEN.
>>> The oentry for cngoto is:
>>>
>>>  { "cngoto", S(CGOTO),0,   3,      "",     "Bl",   ingoto, kngoto, NULL     },
>>>
>>> with 3 as thread, which is OK for THEN. KTHEN, but not ITHEN.
>>>
>>> Should this be cigoto for ITHEN?
>>>
>>>
>>> On 7 Aug 2013, at 12:48, Steven Yi wrote:
>>>
>>>> I think if it's emitting a kngoto, then the relevant i-time opcode
>>>> would be cngoto. One thought I had was that maybe the semantic
>>>> analyzer should check the rates of opcodes within the branch, and
>>>> really determine which to use depending on if any of the opcodes are
>>>> run at perf-time or not.  That seems like it would solve a lot of
>>>> Csound coding issues.  Thoughts?
>>>>
>>>> On Wed, Aug 7, 2013 at 1:15 PM, Victor Lazzarini
>>>>  wrote:
>>>>> As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
>>>>> a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
>>>>> cigoto? Anyway, it should be a simple fix.
>>>>>
>>>>> Dr Victor Lazzarini
>>>>> Senior Lecturer
>>>>> Dept. of Music
>>>>> NUI Maynooth Ireland
>>>>> tel.: +353 1 708 3545
>>>>> Victor dot Lazzarini AT nuim dot ie
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------------
>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>> It's a free troubleshooting tool designed for production.
>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>> Download for free and get started troubleshooting in minutes.
>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>> _______________________________________________
>>>>> Csound-devel mailing list
>>>>> Csound-devel@lists.sourceforge.net
>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>
>>>> ------------------------------------------------------------------------------
>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>> It's a free troubleshooting tool designed for production.
>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>> Download for free and get started troubleshooting in minutes.
>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>> _______________________________________________
>>>> Csound-devel mailing list
>>>> Csound-devel@lists.sourceforge.net
>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>
>>> Dr Victor Lazzarini
>>> Senior Lecturer
>>> Dept. of Music
>>> NUI Maynooth Ireland
>>> tel.: +353 1 708 3545
>>> Victor dot Lazzarini AT nuim dot ie
>>>
>>>
>>>
>>>
>>> ------------------------------------------------------------------------------
>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>> It's a free troubleshooting tool designed for production.
>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>> Download for free and get started troubleshooting in minutes.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> Csound-devel mailing list
>>> Csound-devel@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>
>> ------------------------------------------------------------------------------
>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>> It's a free troubleshooting tool designed for production.
>> Get down to code-level detail for bottlenecks, with <2% overhead.
>> Download for free and get started troubleshooting in minutes.
>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>
>
>
> --
>
> Oeyvind Brandtsegg
> Professor of Music Technology
> NTNU
> 7491 Trondheim
> Norway
> Cell: +47 92 203 205
>
> http://flyndresang.no/
> http://www.partikkelaudio.com/
> http://soundcloud.com/brandtsegg
> http://soundcloud.com/t-emp
>
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

Ok. Yes, I recognize the problem. The gotos are indeed and
"interesting" field full of happy opportunities to shoot oneself in
the foot.
I do appreciate you looking into it and trying to simplify. It just
looked as if changing it could be something that might change the
behaviour of old orchestras.
all best
Oeyvind

2013/8/8 Steven Yi :
> Hi Oeyvind,
>
> Nope, you got it exactly right.  The intention is to change depending
> on what is inside the block. My rationale behind it is as follows:
>
> If's work by jumping over blocks of opcodes.  However, with the case
> of something like:
>
>
> if (ival > 0) ithen
>     krateval someOpcode ival, 0
> else
>     krateval someOpcode ival, 1
> endif
>
> what happens is that you get something like this:
>
> #b0 lt ival, 0
> cingoto #b0, __synth_label0
> krateval someOpcode ival, 0
> igoto __synth_label1
> :__synth_label0
> krateval someOpcode ival, 1
> :__synth_label1
>
> Now, if the ival happens to be greater than zero, you'll get what
> seems like the right behavior. During init-time, the (ival > 0) is
> evaluated and stored into #b0.  cingoto checks and decides not to
> jump, then the first someOpcode is initialized.  Then the goto will
> jump past the rest of the opcodes to synth_label1, and the 2nd
> someOpcode is not initialized.  At perf-time, everything *seems* to be
> fine.  What happens is cingoto has no perf function, so it will pass
> through to the first branch *always*.  The first branch is run and the
> rest is jumped over with the goto.  However, all seems fine.
>
> Now, what happens if ival is <= 0?  The cingoto will evaluate and
> because it is not true, jump to synth_label0.  The 2nd someOpcode now
> is initialized, and the first is not.
>
> Now at perf time, we're in trouble.  Because cingoto has no perf-time
> function, evaluation falls through and we will always go through the
> first branch of code.  The first branch has not be initialized, and we
> get bad behavior. In the case of the attached test, we get a crash
> with a SEGFAULT.
>
>
> Now, my analysis is this: if you have an i-time boolean evaluation
> with an if-then, when you have opcodes that perform at k-rate, it can
> and will cause errors.  Up until now, we seem to have gotten around
> those errors by just always running at k-rate.  So if you wrote code
> with i-time boolean and k-rate opcodes, that was not going to cause
> issues.  However, with i-time only evaluation, we have the
> possibilities for serious problems, as illustrated above and in the
> example CSD.
>
> Note, opcodes that take in k-var's and output k-var's may not
> necessarily be k-rate opcodes.  For example, i(kval) is one, and
> chnexport.
>
> steven
>
>
> On Thu, Aug 8, 2013 at 8:47 AM, Oeyvind Brandtsegg
>  wrote:
>> I'm not sure I understand 100% the implications of this, but are you
>> thinking of changing how the goto operates based on what is inside the
>> relevant (goto'ed) section? I would think it possible (intentionally)
>> that one create an igoto block that may have k-rate opcodes in it.
>> Sorry for the noise if I misunderstood the intention of the change.
>> best
>> Oeyvind
>>
>> 2013/8/8 Steven Yi :
>>> I just took a look.  cigoto won't do it as it's looking for the
>>> positive case, while cngoto check for the negative case.  We'd need to
>>> modify or add an opcode that runs just ingoto and no kngoto. I think
>>> ultimately we should do as mentioned and have the semantic analyzer
>>> choose the correct option depending on if the opcodes within the
>>> branch are i-time only or not.  That will be some work though, due to
>>> nested if's.
>>>
>>> On Wed, Aug 7, 2013 at 2:40 PM, Victor Lazzarini
>>>  wrote:
>>>> The relevant code seems to be in csound_orc_expressions.c:
>>>>
>>>> TREE * create_goto_token(CSOUND *csound, char * booleanVar,
>>>>                          TREE * gotoNode, int type)
>>>> {
>>>> /*     TREE *ans = create_empty_token(csound); */
>>>>     char* op = (char *)csound->Malloc(csound, 7); /* Unchecked */
>>>>     TREE *opTree, *bVar;
>>>>
>>>>     switch(gotoNode->type) {
>>>>     case KGOTO_TOKEN:
>>>>       sprintf(op, "ckgoto");
>>>>       break;
>>>>     case IGOTO_TOKEN:
>>>>       sprintf(op, "cigoto");
>>>>       break;
>>>>     case ITHEN_TOKEN:
>>>>       sprintf(op, "cngoto");
>>>>       break;
>>>>     case THEN_TOKEN:
>>>>     case KTHEN_TOKEN:
>>>>       sprintf(op, "cngoto");
>>>>       break;
>>>>     default:
>>>>       if (type) sprintf(op, "ckgoto");
>>>>       else sprintf(op, "cggoto");
>>>>     }
>>>>
>>>>
>>>> you see that there is no difference between ITHEN, THEN or KTHEN.
>>>> The oentry for cngoto is:
>>>>
>>>>  { "cngoto", S(CGOTO),0,   3,      "",     "Bl",   ingoto, kngoto, NULL     },
>>>>
>>>> with 3 as thread, which is OK for THEN. KTHEN, but not ITHEN.
>>>>
>>>> Should this be cigoto for ITHEN?
>>>>
>>>>
>>>> On 7 Aug 2013, at 12:48, Steven Yi wrote:
>>>>
>>>>> I think if it's emitting a kngoto, then the relevant i-time opcode
>>>>> would be cngoto. One thought I had was that maybe the semantic
>>>>> analyzer should check the rates of opcodes within the branch, and
>>>>> really determine which to use depending on if any of the opcodes are
>>>>> run at perf-time or not.  That seems like it would solve a lot of
>>>>> Csound coding issues.  Thoughts?
>>>>>
>>>>> On Wed, Aug 7, 2013 at 1:15 PM, Victor Lazzarini
>>>>>  wrote:
>>>>>> As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
>>>>>> a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
>>>>>> cigoto? Anyway, it should be a simple fix.
>>>>>>
>>>>>> Dr Victor Lazzarini
>>>>>> Senior Lecturer
>>>>>> Dept. of Music
>>>>>> NUI Maynooth Ireland
>>>>>> tel.: +353 1 708 3545
>>>>>> Victor dot Lazzarini AT nuim dot ie
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> ------------------------------------------------------------------------------
>>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>>> It's a free troubleshooting tool designed for production.
>>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>>> Download for free and get started troubleshooting in minutes.
>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>>> _______________________________________________
>>>>>> Csound-devel mailing list
>>>>>> Csound-devel@lists.sourceforge.net
>>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>>
>>>>> ------------------------------------------------------------------------------
>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>> It's a free troubleshooting tool designed for production.
>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>> Download for free and get started troubleshooting in minutes.
>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>> _______________________________________________
>>>>> Csound-devel mailing list
>>>>> Csound-devel@lists.sourceforge.net
>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>
>>>> Dr Victor Lazzarini
>>>> Senior Lecturer
>>>> Dept. of Music
>>>> NUI Maynooth Ireland
>>>> tel.: +353 1 708 3545
>>>> Victor dot Lazzarini AT nuim dot ie
>>>>
>>>>
>>>>
>>>>
>>>> ------------------------------------------------------------------------------
>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>> It's a free troubleshooting tool designed for production.
>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>> Download for free and get started troubleshooting in minutes.
>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>> _______________________________________________
>>>> Csound-devel mailing list
>>>> Csound-devel@lists.sourceforge.net
>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>
>>> ------------------------------------------------------------------------------
>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>> It's a free troubleshooting tool designed for production.
>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>> Download for free and get started troubleshooting in minutes.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> Csound-devel mailing list
>>> Csound-devel@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>
>>
>>
>> --
>>
>> Oeyvind Brandtsegg
>> Professor of Music Technology
>> NTNU
>> 7491 Trondheim
>> Norway
>> Cell: +47 92 203 205
>>
>> http://flyndresang.no/
>> http://www.partikkelaudio.com/
>> http://soundcloud.com/brandtsegg
>> http://soundcloud.com/t-emp
>>
>> ------------------------------------------------------------------------------
>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>> It's a free troubleshooting tool designed for production.
>> Get down to code-level detail for bottlenecks, with <2% overhead.
>> Download for free and get started troubleshooting in minutes.
>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel
>



-- 

Oeyvind Brandtsegg
Professor of Music Technology
NTNU
7491 Trondheim
Norway
Cell: +47 92 203 205

http://flyndresang.no/
http://www.partikkelaudio.com/
http://soundcloud.com/brandtsegg
http://soundcloud.com/t-emp

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

I think your concern is well-placed.  I think though that this will be
a backwards compatible change.  The logic of the proposed change goes:

1. Current behavior is to always run at i-time and k-time, whether it
needs to run at k-time.
2. Running at k-time for i-time only opcodes does not really do
anything. It just skips over all the i-time opcodes, but wastes some
time.
3. Running at k-time for k-time opcodes prevents crashes/bugs.
4. If analysis occurs and i-time only opcodes found, only run i-branch
at i-time.  If k-time opcodes found, run at k-time as well.

Result is that behavior is maintained.  If all i-time opcodes, runs
only at i-time, if k-time found, run at k-time as well.  It's sort of
an optimization if anything, but it'll also catch bugs in the same way
that always running at i-time and k-time did.

Here's another problem though.  I was looking in the archives and found:

http://csound.1045644.n5.nabble.com/if-then-else-bug-in-csound5-00-td1097353.html

It's from when Csound 5.0 was being worked on.  It's the issue that
all opcodes may need initialization before running.  I'm not sure what
happens for k-rate branches if i-time skipping happens.  (Perhaps it's
time we start working on some developer documentation that specifies
the behavior of the language...)

On Thu, Aug 8, 2013 at 10:10 AM, Oeyvind Brandtsegg
 wrote:
> Ok. Yes, I recognize the problem. The gotos are indeed and
> "interesting" field full of happy opportunities to shoot oneself in
> the foot.
> I do appreciate you looking into it and trying to simplify. It just
> looked as if changing it could be something that might change the
> behaviour of old orchestras.
> all best
> Oeyvind
>
> 2013/8/8 Steven Yi :
>> Hi Oeyvind,
>>
>> Nope, you got it exactly right.  The intention is to change depending
>> on what is inside the block. My rationale behind it is as follows:
>>
>> If's work by jumping over blocks of opcodes.  However, with the case
>> of something like:
>>
>>
>> if (ival > 0) ithen
>>     krateval someOpcode ival, 0
>> else
>>     krateval someOpcode ival, 1
>> endif
>>
>> what happens is that you get something like this:
>>
>> #b0 lt ival, 0
>> cingoto #b0, __synth_label0
>> krateval someOpcode ival, 0
>> igoto __synth_label1
>> :__synth_label0
>> krateval someOpcode ival, 1
>> :__synth_label1
>>
>> Now, if the ival happens to be greater than zero, you'll get what
>> seems like the right behavior. During init-time, the (ival > 0) is
>> evaluated and stored into #b0.  cingoto checks and decides not to
>> jump, then the first someOpcode is initialized.  Then the goto will
>> jump past the rest of the opcodes to synth_label1, and the 2nd
>> someOpcode is not initialized.  At perf-time, everything *seems* to be
>> fine.  What happens is cingoto has no perf function, so it will pass
>> through to the first branch *always*.  The first branch is run and the
>> rest is jumped over with the goto.  However, all seems fine.
>>
>> Now, what happens if ival is <= 0?  The cingoto will evaluate and
>> because it is not true, jump to synth_label0.  The 2nd someOpcode now
>> is initialized, and the first is not.
>>
>> Now at perf time, we're in trouble.  Because cingoto has no perf-time
>> function, evaluation falls through and we will always go through the
>> first branch of code.  The first branch has not be initialized, and we
>> get bad behavior. In the case of the attached test, we get a crash
>> with a SEGFAULT.
>>
>>
>> Now, my analysis is this: if you have an i-time boolean evaluation
>> with an if-then, when you have opcodes that perform at k-rate, it can
>> and will cause errors.  Up until now, we seem to have gotten around
>> those errors by just always running at k-rate.  So if you wrote code
>> with i-time boolean and k-rate opcodes, that was not going to cause
>> issues.  However, with i-time only evaluation, we have the
>> possibilities for serious problems, as illustrated above and in the
>> example CSD.
>>
>> Note, opcodes that take in k-var's and output k-var's may not
>> necessarily be k-rate opcodes.  For example, i(kval) is one, and
>> chnexport.
>>
>> steven
>>
>>
>> On Thu, Aug 8, 2013 at 8:47 AM, Oeyvind Brandtsegg
>>  wrote:
>>> I'm not sure I understand 100% the implications of this, but are you
>>> thinking of changing how the goto operates based on what is inside the
>>> relevant (goto'ed) section? I would think it possible (intentionally)
>>> that one create an igoto block that may have k-rate opcodes in it.
>>> Sorry for the noise if I misunderstood the intention of the change.
>>> best
>>> Oeyvind
>>>
>>> 2013/8/8 Steven Yi :
>>>> I just took a look.  cigoto won't do it as it's looking for the
>>>> positive case, while cngoto check for the negative case.  We'd need to
>>>> modify or add an opcode that runs just ingoto and no kngoto. I think
>>>> ultimately we should do as mentioned and have the semantic analyzer
>>>> choose the correct option depending on if the opcodes within the
>>>> branch are i-time only or not.  That will be some work though, due to
>>>> nested if's.
>>>>
>>>> On Wed, Aug 7, 2013 at 2:40 PM, Victor Lazzarini
>>>>  wrote:
>>>>> The relevant code seems to be in csound_orc_expressions.c:
>>>>>
>>>>> TREE * create_goto_token(CSOUND *csound, char * booleanVar,
>>>>>                          TREE * gotoNode, int type)
>>>>> {
>>>>> /*     TREE *ans = create_empty_token(csound); */
>>>>>     char* op = (char *)csound->Malloc(csound, 7); /* Unchecked */
>>>>>     TREE *opTree, *bVar;
>>>>>
>>>>>     switch(gotoNode->type) {
>>>>>     case KGOTO_TOKEN:
>>>>>       sprintf(op, "ckgoto");
>>>>>       break;
>>>>>     case IGOTO_TOKEN:
>>>>>       sprintf(op, "cigoto");
>>>>>       break;
>>>>>     case ITHEN_TOKEN:
>>>>>       sprintf(op, "cngoto");
>>>>>       break;
>>>>>     case THEN_TOKEN:
>>>>>     case KTHEN_TOKEN:
>>>>>       sprintf(op, "cngoto");
>>>>>       break;
>>>>>     default:
>>>>>       if (type) sprintf(op, "ckgoto");
>>>>>       else sprintf(op, "cggoto");
>>>>>     }
>>>>>
>>>>>
>>>>> you see that there is no difference between ITHEN, THEN or KTHEN.
>>>>> The oentry for cngoto is:
>>>>>
>>>>>  { "cngoto", S(CGOTO),0,   3,      "",     "Bl",   ingoto, kngoto, NULL     },
>>>>>
>>>>> with 3 as thread, which is OK for THEN. KTHEN, but not ITHEN.
>>>>>
>>>>> Should this be cigoto for ITHEN?
>>>>>
>>>>>
>>>>> On 7 Aug 2013, at 12:48, Steven Yi wrote:
>>>>>
>>>>>> I think if it's emitting a kngoto, then the relevant i-time opcode
>>>>>> would be cngoto. One thought I had was that maybe the semantic
>>>>>> analyzer should check the rates of opcodes within the branch, and
>>>>>> really determine which to use depending on if any of the opcodes are
>>>>>> run at perf-time or not.  That seems like it would solve a lot of
>>>>>> Csound coding issues.  Thoughts?
>>>>>>
>>>>>> On Wed, Aug 7, 2013 at 1:15 PM, Victor Lazzarini
>>>>>>  wrote:
>>>>>>> As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
>>>>>>> a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
>>>>>>> cigoto? Anyway, it should be a simple fix.
>>>>>>>
>>>>>>> Dr Victor Lazzarini
>>>>>>> Senior Lecturer
>>>>>>> Dept. of Music
>>>>>>> NUI Maynooth Ireland
>>>>>>> tel.: +353 1 708 3545
>>>>>>> Victor dot Lazzarini AT nuim dot ie
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> ------------------------------------------------------------------------------
>>>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>>>> It's a free troubleshooting tool designed for production.
>>>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>>>> Download for free and get started troubleshooting in minutes.
>>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>>>> _______________________________________________
>>>>>>> Csound-devel mailing list
>>>>>>> Csound-devel@lists.sourceforge.net
>>>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>>>
>>>>>> ------------------------------------------------------------------------------
>>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>>> It's a free troubleshooting tool designed for production.
>>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>>> Download for free and get started troubleshooting in minutes.
>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>>> _______________________________________________
>>>>>> Csound-devel mailing list
>>>>>> Csound-devel@lists.sourceforge.net
>>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>>
>>>>> Dr Victor Lazzarini
>>>>> Senior Lecturer
>>>>> Dept. of Music
>>>>> NUI Maynooth Ireland
>>>>> tel.: +353 1 708 3545
>>>>> Victor dot Lazzarini AT nuim dot ie
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> ------------------------------------------------------------------------------
>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>> It's a free troubleshooting tool designed for production.
>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>> Download for free and get started troubleshooting in minutes.
>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>> _______________________________________________
>>>>> Csound-devel mailing list
>>>>> Csound-devel@lists.sourceforge.net
>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>
>>>> ------------------------------------------------------------------------------
>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>> It's a free troubleshooting tool designed for production.
>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>> Download for free and get started troubleshooting in minutes.
>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>> _______________________________________________
>>>> Csound-devel mailing list
>>>> Csound-devel@lists.sourceforge.net
>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>
>>>
>>>
>>> --
>>>
>>> Oeyvind Brandtsegg
>>> Professor of Music Technology
>>> NTNU
>>> 7491 Trondheim
>>> Norway
>>> Cell: +47 92 203 205
>>>
>>> http://flyndresang.no/
>>> http://www.partikkelaudio.com/
>>> http://soundcloud.com/brandtsegg
>>> http://soundcloud.com/t-emp
>>>
>>> ------------------------------------------------------------------------------
>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>> It's a free troubleshooting tool designed for production.
>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>> Download for free and get started troubleshooting in minutes.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> Csound-devel mailing list
>>> Csound-devel@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>
>> ------------------------------------------------------------------------------
>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>> It's a free troubleshooting tool designed for production.
>> Get down to code-level detail for bottlenecks, with <2% overhead.
>> Download for free and get started troubleshooting in minutes.
>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>
>
>
>
> --
>
> Oeyvind Brandtsegg
> Professor of Music Technology
> NTNU
> 7491 Trondheim
> Norway
> Cell: +47 92 203 205
>
> http://flyndresang.no/
> http://www.partikkelaudio.com/
> http://soundcloud.com/brandtsegg
> http://soundcloud.com/t-emp
>
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

All good,
thanks for the explanation.
best
Oeyvind

2013/8/8 Steven Yi :
> I think your concern is well-placed.  I think though that this will be
> a backwards compatible change.  The logic of the proposed change goes:
>
> 1. Current behavior is to always run at i-time and k-time, whether it
> needs to run at k-time.
> 2. Running at k-time for i-time only opcodes does not really do
> anything. It just skips over all the i-time opcodes, but wastes some
> time.
> 3. Running at k-time for k-time opcodes prevents crashes/bugs.
> 4. If analysis occurs and i-time only opcodes found, only run i-branch
> at i-time.  If k-time opcodes found, run at k-time as well.
>
> Result is that behavior is maintained.  If all i-time opcodes, runs
> only at i-time, if k-time found, run at k-time as well.  It's sort of
> an optimization if anything, but it'll also catch bugs in the same way
> that always running at i-time and k-time did.
>
> Here's another problem though.  I was looking in the archives and found:
>
> http://csound.1045644.n5.nabble.com/if-then-else-bug-in-csound5-00-td1097353.html
>
> It's from when Csound 5.0 was being worked on.  It's the issue that
> all opcodes may need initialization before running.  I'm not sure what
> happens for k-rate branches if i-time skipping happens.  (Perhaps it's
> time we start working on some developer documentation that specifies
> the behavior of the language...)
>
> On Thu, Aug 8, 2013 at 10:10 AM, Oeyvind Brandtsegg
>  wrote:
>> Ok. Yes, I recognize the problem. The gotos are indeed and
>> "interesting" field full of happy opportunities to shoot oneself in
>> the foot.
>> I do appreciate you looking into it and trying to simplify. It just
>> looked as if changing it could be something that might change the
>> behaviour of old orchestras.
>> all best
>> Oeyvind
>>
>> 2013/8/8 Steven Yi :
>>> Hi Oeyvind,
>>>
>>> Nope, you got it exactly right.  The intention is to change depending
>>> on what is inside the block. My rationale behind it is as follows:
>>>
>>> If's work by jumping over blocks of opcodes.  However, with the case
>>> of something like:
>>>
>>>
>>> if (ival > 0) ithen
>>>     krateval someOpcode ival, 0
>>> else
>>>     krateval someOpcode ival, 1
>>> endif
>>>
>>> what happens is that you get something like this:
>>>
>>> #b0 lt ival, 0
>>> cingoto #b0, __synth_label0
>>> krateval someOpcode ival, 0
>>> igoto __synth_label1
>>> :__synth_label0
>>> krateval someOpcode ival, 1
>>> :__synth_label1
>>>
>>> Now, if the ival happens to be greater than zero, you'll get what
>>> seems like the right behavior. During init-time, the (ival > 0) is
>>> evaluated and stored into #b0.  cingoto checks and decides not to
>>> jump, then the first someOpcode is initialized.  Then the goto will
>>> jump past the rest of the opcodes to synth_label1, and the 2nd
>>> someOpcode is not initialized.  At perf-time, everything *seems* to be
>>> fine.  What happens is cingoto has no perf function, so it will pass
>>> through to the first branch *always*.  The first branch is run and the
>>> rest is jumped over with the goto.  However, all seems fine.
>>>
>>> Now, what happens if ival is <= 0?  The cingoto will evaluate and
>>> because it is not true, jump to synth_label0.  The 2nd someOpcode now
>>> is initialized, and the first is not.
>>>
>>> Now at perf time, we're in trouble.  Because cingoto has no perf-time
>>> function, evaluation falls through and we will always go through the
>>> first branch of code.  The first branch has not be initialized, and we
>>> get bad behavior. In the case of the attached test, we get a crash
>>> with a SEGFAULT.
>>>
>>>
>>> Now, my analysis is this: if you have an i-time boolean evaluation
>>> with an if-then, when you have opcodes that perform at k-rate, it can
>>> and will cause errors.  Up until now, we seem to have gotten around
>>> those errors by just always running at k-rate.  So if you wrote code
>>> with i-time boolean and k-rate opcodes, that was not going to cause
>>> issues.  However, with i-time only evaluation, we have the
>>> possibilities for serious problems, as illustrated above and in the
>>> example CSD.
>>>
>>> Note, opcodes that take in k-var's and output k-var's may not
>>> necessarily be k-rate opcodes.  For example, i(kval) is one, and
>>> chnexport.
>>>
>>> steven
>>>
>>>
>>> On Thu, Aug 8, 2013 at 8:47 AM, Oeyvind Brandtsegg
>>>  wrote:
>>>> I'm not sure I understand 100% the implications of this, but are you
>>>> thinking of changing how the goto operates based on what is inside the
>>>> relevant (goto'ed) section? I would think it possible (intentionally)
>>>> that one create an igoto block that may have k-rate opcodes in it.
>>>> Sorry for the noise if I misunderstood the intention of the change.
>>>> best
>>>> Oeyvind
>>>>
>>>> 2013/8/8 Steven Yi :
>>>>> I just took a look.  cigoto won't do it as it's looking for the
>>>>> positive case, while cngoto check for the negative case.  We'd need to
>>>>> modify or add an opcode that runs just ingoto and no kngoto. I think
>>>>> ultimately we should do as mentioned and have the semantic analyzer
>>>>> choose the correct option depending on if the opcodes within the
>>>>> branch are i-time only or not.  That will be some work though, due to
>>>>> nested if's.
>>>>>
>>>>> On Wed, Aug 7, 2013 at 2:40 PM, Victor Lazzarini
>>>>>  wrote:
>>>>>> The relevant code seems to be in csound_orc_expressions.c:
>>>>>>
>>>>>> TREE * create_goto_token(CSOUND *csound, char * booleanVar,
>>>>>>                          TREE * gotoNode, int type)
>>>>>> {
>>>>>> /*     TREE *ans = create_empty_token(csound); */
>>>>>>     char* op = (char *)csound->Malloc(csound, 7); /* Unchecked */
>>>>>>     TREE *opTree, *bVar;
>>>>>>
>>>>>>     switch(gotoNode->type) {
>>>>>>     case KGOTO_TOKEN:
>>>>>>       sprintf(op, "ckgoto");
>>>>>>       break;
>>>>>>     case IGOTO_TOKEN:
>>>>>>       sprintf(op, "cigoto");
>>>>>>       break;
>>>>>>     case ITHEN_TOKEN:
>>>>>>       sprintf(op, "cngoto");
>>>>>>       break;
>>>>>>     case THEN_TOKEN:
>>>>>>     case KTHEN_TOKEN:
>>>>>>       sprintf(op, "cngoto");
>>>>>>       break;
>>>>>>     default:
>>>>>>       if (type) sprintf(op, "ckgoto");
>>>>>>       else sprintf(op, "cggoto");
>>>>>>     }
>>>>>>
>>>>>>
>>>>>> you see that there is no difference between ITHEN, THEN or KTHEN.
>>>>>> The oentry for cngoto is:
>>>>>>
>>>>>>  { "cngoto", S(CGOTO),0,   3,      "",     "Bl",   ingoto, kngoto, NULL     },
>>>>>>
>>>>>> with 3 as thread, which is OK for THEN. KTHEN, but not ITHEN.
>>>>>>
>>>>>> Should this be cigoto for ITHEN?
>>>>>>
>>>>>>
>>>>>> On 7 Aug 2013, at 12:48, Steven Yi wrote:
>>>>>>
>>>>>>> I think if it's emitting a kngoto, then the relevant i-time opcode
>>>>>>> would be cngoto. One thought I had was that maybe the semantic
>>>>>>> analyzer should check the rates of opcodes within the branch, and
>>>>>>> really determine which to use depending on if any of the opcodes are
>>>>>>> run at perf-time or not.  That seems like it would solve a lot of
>>>>>>> Csound coding issues.  Thoughts?
>>>>>>>
>>>>>>> On Wed, Aug 7, 2013 at 1:15 PM, Victor Lazzarini
>>>>>>>  wrote:
>>>>>>>> As I mentioned yesterday, the if ... ithen statement is being parsed into the wrong opcode, kngoto. This is
>>>>>>>> a bug that has been there since the Csound 5. I am not quite sure which opcode it should use, is it
>>>>>>>> cigoto? Anyway, it should be a simple fix.
>>>>>>>>
>>>>>>>> Dr Victor Lazzarini
>>>>>>>> Senior Lecturer
>>>>>>>> Dept. of Music
>>>>>>>> NUI Maynooth Ireland
>>>>>>>> tel.: +353 1 708 3545
>>>>>>>> Victor dot Lazzarini AT nuim dot ie
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> ------------------------------------------------------------------------------
>>>>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>>>>> It's a free troubleshooting tool designed for production.
>>>>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>>>>> Download for free and get started troubleshooting in minutes.
>>>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>>>>> _______________________________________________
>>>>>>>> Csound-devel mailing list
>>>>>>>> Csound-devel@lists.sourceforge.net
>>>>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>>>>
>>>>>>> ------------------------------------------------------------------------------
>>>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>>>> It's a free troubleshooting tool designed for production.
>>>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>>>> Download for free and get started troubleshooting in minutes.
>>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>>>> _______________________________________________
>>>>>>> Csound-devel mailing list
>>>>>>> Csound-devel@lists.sourceforge.net
>>>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>>>
>>>>>> Dr Victor Lazzarini
>>>>>> Senior Lecturer
>>>>>> Dept. of Music
>>>>>> NUI Maynooth Ireland
>>>>>> tel.: +353 1 708 3545
>>>>>> Victor dot Lazzarini AT nuim dot ie
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> ------------------------------------------------------------------------------
>>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>>> It's a free troubleshooting tool designed for production.
>>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>>> Download for free and get started troubleshooting in minutes.
>>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>>> _______________________________________________
>>>>>> Csound-devel mailing list
>>>>>> Csound-devel@lists.sourceforge.net
>>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>>
>>>>> ------------------------------------------------------------------------------
>>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>>> It's a free troubleshooting tool designed for production.
>>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>>> Download for free and get started troubleshooting in minutes.
>>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>>> _______________________________________________
>>>>> Csound-devel mailing list
>>>>> Csound-devel@lists.sourceforge.net
>>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>>
>>>>
>>>>
>>>> --
>>>>
>>>> Oeyvind Brandtsegg
>>>> Professor of Music Technology
>>>> NTNU
>>>> 7491 Trondheim
>>>> Norway
>>>> Cell: +47 92 203 205
>>>>
>>>> http://flyndresang.no/
>>>> http://www.partikkelaudio.com/
>>>> http://soundcloud.com/brandtsegg
>>>> http://soundcloud.com/t-emp
>>>>
>>>> ------------------------------------------------------------------------------
>>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>>> It's a free troubleshooting tool designed for production.
>>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>>> Download for free and get started troubleshooting in minutes.
>>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>>> _______________________________________________
>>>> Csound-devel mailing list
>>>> Csound-devel@lists.sourceforge.net
>>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>
>>> ------------------------------------------------------------------------------
>>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>>> It's a free troubleshooting tool designed for production.
>>> Get down to code-level detail for bottlenecks, with <2% overhead.
>>> Download for free and get started troubleshooting in minutes.
>>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>>> _______________________________________________
>>> Csound-devel mailing list
>>> Csound-devel@lists.sourceforge.net
>>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>>
>>
>>
>>
>> --
>>
>> Oeyvind Brandtsegg
>> Professor of Music Technology
>> NTNU
>> 7491 Trondheim
>> Norway
>> Cell: +47 92 203 205
>>
>> http://flyndresang.no/
>> http://www.partikkelaudio.com/
>> http://soundcloud.com/brandtsegg
>> http://soundcloud.com/t-emp
>>
>> ------------------------------------------------------------------------------
>> Get 100% visibility into Java/.NET code with AppDynamics Lite!
>> It's a free troubleshooting tool designed for production.
>> Get down to code-level detail for bottlenecks, with <2% overhead.
>> Download for free and get started troubleshooting in minutes.
>> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel



-- 

Oeyvind Brandtsegg
Professor of Music Technology
NTNU
7491 Trondheim
Norway
Cell: +47 92 203 205

http://flyndresang.no/
http://www.partikkelaudio.com/
http://soundcloud.com/brandtsegg
http://soundcloud.com/t-emp

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net