Re: CET 20-01-01 Adobe way of growing the dictionary |

Alex,

Did you happen to check if there was any performance difference (I don't 
expect
anything significant) ?

I see no reason not to switch to Adobe's method. It fixes a CET issue 
and doesn't
seem like an unreasonable approach.

Regards,
Ray
------------------------------------------------------------------------

Alex Cherepanov wrote:
> Reproduce the algorithm used by Adobe to grow the dictionary
> to match CET 20-01-01.
>
> DIFFERENCES:
> None: no comparefiles or CET differences.
>
> The main question for the reviewer: do we need this adjustment
> at all ?
>
> ------------------------------------------------------------------------
>
> Index: gs/src/idict.c
> ===================================================================
> --- gs/src/idict.c	(revision 7709)
> +++ gs/src/idict.c	(working copy)
> @@ -790,8 +790,14 @@
>      dict *pdict = pdref->value.pdict;
>      /* We might have maxlength < npairs, if */
>      /* dict_round_size increased the size. */
> -    ulong new_size = (ulong) d_maxlength(pdict) * 3 / 2 + 2;
> -
> +    ulong new_size = (ulong) d_maxlength(pdict);
> +    /* Adobe does this */
> +    if (new_size < 20)       
> +        new_size += 10;
> +    else if (new_size < 200)
> +        new_size *= 2;
> +    else
> +        new_size += new_size / 2;
>  #if arch_sizeof_int < arch_sizeof_long
>      if (new_size > max_uint)
>  	new_size = max_uint;
>
>   
> ------------------------------------------------------------------------
>
> _______________________________________________
> gs-code-review mailing list
> [email protected]
> http://www.ghostscript.com/mailman/listinfo/gs-code-review
>