COSC_4P82_Assignment_1/lib/lilgp/peters-patch.html

<HTML>
<HEAD>
<TITLE>
Peter's Patch
</TITLE>
</HEAD>
<BODY bgcolor=white>
<H1>Peter's Patch to Sean's Patched <i>lil-gp</i>  <a href="http://www.cs.umd.edu/users/seanl/patched-gp/gp.tar.gz">Kernel</a></h1>

<p><a href="http://www.daimi.aau.dk/~ptr/">Peter Anderson</a> noted the following bug in the ERC facility.  I've not yet verified that this bug exists or that the patch works, so you're own your own.  I include his message to me below:

<pre>

Use of ERC terminal on the form "name:ERCindex" do not work.

The ECR terminal nodes are not constructed.
The first call to "mod_read_tree_recurse" 
eind with the ERC structure is NULL, and the call to
"get_function_by_name" core dump on a I:1 structure.

I have modified the function "mod_read_tree_recurse" in populate.c so
the user can use ERC integer constant. By supplying a function to
insert the constant in the DATATYPE structure. The chances are as followed.
in the files:

populate.c
ckpoint.c
 

a) "get_function_by_name" in ckpoint.c is made safe by changing:
  
        *ep = eind[j];
        (*ep)->f = fs->cset+i;

change to:

        if (eind != NULL) {
            *ep = eind[j];
            (*ep)->f = fs->cset+i;
        }

b) "mod_read_tree_recurs" catch the ERCindex before the call to 
"get_function_by_name"

  f = get_function_by_name ( tree, string, &ep, eind );

chance to:

  sl = strlen ( string );
  for ( si = 0; si < sl; ++si )
    {

      if ( string[si] == ':' )
        {
          /* names of the form "name:index" are chopped at the colon,
             and the value of the index saved. */
          string[si] = 0;
          ECRvalue = atoi ( string+si+1 );
          break;
        }
    }

  f = get_function_by_name ( tree, string, &ep, eind );

where sl,si are int.

c) create the ERC'c in "mod_read_tree_recurse"

  switch ( f->type )
    {
    case TERM_NORM:
    case TERM_ARG:
    case EVAL_TERM:
      break;
    case TERM_ERC:
      /* stt_record the ERC address as the next lnode in the array. */
      gensp_next(space)->d = ep; 
      break;

chance to:

  switch ( f->type )
    {
    case TERM_NORM:
    case TERM_ARG:
    case EVAL_TERM:
      break;
    case TERM_ERC:
      /* stt_record the ERC address as the next lnode in the array. */
      new_ECR = gensp_next(space);
      new_ECR->d = new_ephemeral_const ( f );
      set_ECR_value(&new_ECR->d->d , ECRvalue);
      /* gensp_next(space)->d = ep; */
      break;

The function "set_ECR_value(&new_ECR->d->d , ECRvalue);" is user defined
in my case it is:

void set_ECR_value(DATATYPE *d, int ECRvalue)
{
  d->input = ECRvalue;
}


d) before the first call to "mod_read_tree_recurse" is there a call to
"gensp_reset(0)" that is not needed.

  case LOAD_FILE:
    gensp_reset(0); 
    mod_read_tree_recurse( 0, NULL, dataum[j], j, buf, 1);
    /* need to do something here about limit checking */
    break;

chance to:

  case LOAD_FILE:
    /* gensp_reset(0); not needed there is a call before the switch */
    mod_read_tree_recurse( 0, NULL, dataum[j], j, buf, 1);
    /* need to do something here about limit checking */
    break;


</pre>
init 2024-02-13 21:21:51 -05:00			`<HTML>`
			`<HEAD>`
			`<TITLE>`
			`Peter's Patch`
			`</TITLE>`
			`</HEAD>`
			`<BODY bgcolor=white>`
			`<H1>Peter's Patch to Sean's Patched <i>lil-gp</i> <a href="http://www.cs.umd.edu/users/seanl/patched-gp/gp.tar.gz">Kernel</a></h1>`

			`<p><a href="http://www.daimi.aau.dk/~ptr/">Peter Anderson</a> noted the following bug in the ERC facility. I've not yet verified that this bug exists or that the patch works, so you're own your own. I include his message to me below:`

			`<pre>`

			`Use of ERC terminal on the form "name:ERCindex" do not work.`

			`The ECR terminal nodes are not constructed.`
			`The first call to "mod_read_tree_recurse"`
			`eind with the ERC structure is NULL, and the call to`
			`"get_function_by_name" core dump on a I:1 structure.`

			`I have modified the function "mod_read_tree_recurse" in populate.c so`
			`the user can use ERC integer constant. By supplying a function to`
			`insert the constant in the DATATYPE structure. The chances are as followed.`
			`in the files:`

			`populate.c`
			`ckpoint.c`


			`a) "get_function_by_name" in ckpoint.c is made safe by changing:`

			`*ep = eind[j];`
			`(*ep)->f = fs->cset+i;`

			`change to:`

			`if (eind != NULL) {`
			`*ep = eind[j];`
			`(*ep)->f = fs->cset+i;`
			`}`

			`b) "mod_read_tree_recurs" catch the ERCindex before the call to`
			`"get_function_by_name"`

			`f = get_function_by_name ( tree, string, &ep, eind );`

			`chance to:`

			`sl = strlen ( string );`
			`for ( si = 0; si < sl; ++si )`
			`{`

			`if ( string[si] == ':' )`
			`{`
			`/* names of the form "name:index" are chopped at the colon,`
			`and the value of the index saved. */`
			`string[si] = 0;`
			`ECRvalue = atoi ( string+si+1 );`
			`break;`
			`}`
			`}`

			`f = get_function_by_name ( tree, string, &ep, eind );`

			`where sl,si are int.`

			`c) create the ERC'c in "mod_read_tree_recurse"`

			`switch ( f->type )`
			`{`
			`case TERM_NORM:`
			`case TERM_ARG:`
			`case EVAL_TERM:`
			`break;`
			`case TERM_ERC:`
sexy 2024-02-16 00:05:14 -05:00			`/* stt_record the ERC address as the next lnode in the array. */`
init 2024-02-13 21:21:51 -05:00			`gensp_next(space)->d = ep;`
			`break;`

			`chance to:`

			`switch ( f->type )`
			`{`
			`case TERM_NORM:`
			`case TERM_ARG:`
			`case EVAL_TERM:`
			`break;`
			`case TERM_ERC:`
sexy 2024-02-16 00:05:14 -05:00			`/* stt_record the ERC address as the next lnode in the array. */`
init 2024-02-13 21:21:51 -05:00			`new_ECR = gensp_next(space);`
			`new_ECR->d = new_ephemeral_const ( f );`
			`set_ECR_value(&new_ECR->d->d , ECRvalue);`
			`/* gensp_next(space)->d = ep; */`
			`break;`

			`The function "set_ECR_value(&new_ECR->d->d , ECRvalue);" is user defined`
			`in my case it is:`

			`void set_ECR_value(DATATYPE *d, int ECRvalue)`
			`{`
			`d->input = ECRvalue;`
			`}`


			`d) before the first call to "mod_read_tree_recurse" is there a call to`
			`"gensp_reset(0)" that is not needed.`

			`case LOAD_FILE:`
			`gensp_reset(0);`
			`mod_read_tree_recurse( 0, NULL, dataum[j], j, buf, 1);`
			`/* need to do something here about limit checking */`
			`break;`

			`chance to:`

			`case LOAD_FILE:`
			`/* gensp_reset(0); not needed there is a call before the switch */`
			`mod_read_tree_recurse( 0, NULL, dataum[j], j, buf, 1);`
			`/* need to do something here about limit checking */`
			`break;`


			`</pre>`