cobolforgcc-devel

[Cobolforgcc-devel] Sort definition

[Tim J. <te...@me...>]

Ted,

How about this for the basic in core sort? You could do it in one
call but then you would need to reanalyze the sort field details
for each buffer full of records.

Regards,
Tim Josling

/* set up for one or more sorts */

enum cobr_sort_status_enum
cobr_basic_sort_init
(
 UINT32 sort_field_count,
 struct sort_field_desc* sort_fields,
 UINT8* collating_sequence /* 256 bytes in order for text field
sort or NULL */
);

/* do a sort */

enum cobr_sort_status_enum
cobr_basic_sort(
 void * handle,
 UINT32 record_count, 
 void** records, 
 UINT32* record_lengths);

/* free the memory etc */

enum cobr_sort_status_enum
cobr_basic_sort_wrap(void * handle);

struct sort_field_desc
{
UINT32 offset;
UINT32 length;
UINT32 decimals;
UINT32 type;
UINT32 flags;
};

enum cobr_sort_status_enum
{ /* sample only */
cobr_sort_status_OK=0,
cobr_sort_status_invalid_key_definitions,
cobr_sort_status_record_too_short,
...etc...
};

<end>

RE: [Cobolforgcc-devel] Sort definition

[William M. K. <wm...@ix...>]

I seem to have missed whatever note this is replying to - but does this
handle both

   WITH DUPLICATES

	and

   COLLATING SEQUENCE clauses?

Also, what (if anything) does this do with INPUT/OUTPUT procedures?

If you "implement" a simple (Using/Giving - native collating sequence -
only) SORT, then I suggest that you add a compile-time message "warning"
that other phrases are "currently ignored"

P.S.  There is a VERY obscure rule in the ANSI Standard (that some compiler
ignore and most programmers do NOT know about) which states that you "must"
go to a File Declarative in the middle of a SORT - should the "condition"
specified in the DECLARATIVE occur anywhere during the SORT.  Just thought I
would mention this - so you might want to "design" around this.

> -----Original Message-----
> From: cob...@li...
> [mailto:cob...@li...]On Behalf Of Tim
> Josling
> Sent: Tuesday, December 26, 2000 4:28 AM
> To: Tim Josling
> Cc: cobolforgcc-devel
> Subject: [Cobolforgcc-devel] Sort definition
>
>
> Ted,
>
> How about this for the basic in core sort? You could do it in one
> call but then you would need to reanalyze the sort field details
> for each buffer full of records.
>
> Regards,
> Tim Josling
>
> /* set up for one or more sorts */
>
> enum cobr_sort_status_enum
> cobr_basic_sort_init
> (
>  UINT32 sort_field_count,
>  struct sort_field_desc* sort_fields,
>  UINT8* collating_sequence /* 256 bytes in order for text field
> sort or NULL */
> );
>
> /* do a sort */
>
> enum cobr_sort_status_enum
> cobr_basic_sort(
>  void * handle,
>  UINT32 record_count,
>  void** records,
>  UINT32* record_lengths);
>
> /* free the memory etc */
>
> enum cobr_sort_status_enum
> cobr_basic_sort_wrap(void * handle);
>
> struct sort_field_desc
> {
> UINT32 offset;
> UINT32 length;
> UINT32 decimals;
> UINT32 type;
> UINT32 flags;
> };
>
> enum cobr_sort_status_enum
> { /* sample only */
> cobr_sort_status_OK=0,
> cobr_sort_status_invalid_key_definitions,
> cobr_sort_status_record_too_short,
> ...etc...
> };
>
> <end>
>
>
>
> _______________________________________________
> Cobolforgcc-devel mailing list
> Cob...@li...
> http://lists.sourceforge.net/mailman/listinfo/cobolforgcc-devel

Re: [Cobolforgcc-devel] Sort definition

[Tim J. <te...@me...>]

Bill,

Thanks. Which section of the standard are you referring to?
Declaratives are down the track but it is good to know these
things in advance. 

The spec below is just for the 'inner sort' - other moodules will
do the file IO, interfacing to the compiled code, overflowing to
disk and so on. So it is not as bad as it looks. 

Regards,
Tim Josling

PS no need to cc me I always read the mailing list.

"William M. Klein" wrote:
> 
> I seem to have missed whatever note this is replying to - but does this
> handle both
> 
>    WITH DUPLICATES
> 
>         and
> 
>    COLLATING SEQUENCE clauses?
> 
> Also, what (if anything) does this do with INPUT/OUTPUT procedures?
> 
> If you "implement" a simple (Using/Giving - native collating sequence -
> only) SORT, then I suggest that you add a compile-time message "warning"
> that other phrases are "currently ignored"
> 
> P.S.  There is a VERY obscure rule in the ANSI Standard (that some compiler
> ignore and most programmers do NOT know about) which states that you "must"
> go to a File Declarative in the middle of a SORT - should the "condition"
> specified in the DECLARATIVE occur anywhere during the SORT.  Just thought I
> would mention this - so you might want to "design" around this.
>