Re: Modifying update_attstats of analyze.c for C Strings

As a follow-up question,

I found some of the varchar column types, in which the histogram_bounds are
not being surrounded in double quotes (" ") even in the default
implementation.
Ex : *c_name* column of *Customer* table

I also found histogram_bounds in which only some strings are surrounded in
double quotes and some are not.
Ex : *c_address *column of* Customer *table

Why are there such inconsistencies? How is this determined?

Thank you.

On Tue, Jul 8, 2014 at 10:52 AM, Ashoke <s(dot)ashoke(at)gmail(dot)com> wrote:

> Hi,
>
> I am trying to implement a functionality that is similar to ANALYZE, but
> needs to have different values (the values will be valid and is stored in
> inp->str[][]) for MCV/Histogram Bounds in case the column under
> consideration is varchar (C Strings). I have written a function
> *dummy_update_attstats* with the following changes. Other things remain
> the same as in *update_attstats* of *~/src/backend/commands/analyze.c*
>
>
> *---*
> *{*
>
> * ArrayType *arry; *
> * if (*
> *strcmp(col_type,"varchar") == 0*
> * )*
> * arry = construct_array(stats->stavalues[k],*
> * stats->numvalues[k], *
> * CSTRINGOID,*
> * -2, *
> * false,*
> * 'c'); *
> * else*
> * arry = construct_array(stats->stavalues[k], *
> * stats->numvalues[k],*
> * stats->statypid[k], *
> * stats->statyplen[k],*
> * stats->statypbyval[k], *
> * stats->statypalign[k]);*
> * values[i++] = PointerGetDatum(arry); /* stavaluesN */ }*
> ---
>
> and I update the hist_values in the appropriate function as:
> ---
>
> *if (strcmp(col_type,"varchar") == 0**)*
> * hist_values[i] = datumCopy(CStringGetDatum(inp->str[i][j]),*
> * false,*
> * -2);*
> *---*
>
> I tried this based on the following reference :
> http://www.postgresql.org/message-id/attachment/20352/vacattrstats-extend.diff
>
> My issue is : When I use my way for strings, the MCV/histogram_bounds in
> pg_stats doesn't have double quotes (" ") surrounding string. That is,
>
> If normal *update_attstats* is used, histogram_bounds for *TPCH
> nation(n_name)* are : *"ALGERIA ","ARGENTINA ",...*
> If I use *dummy_update_attstats* as above, histogram_bounds for *TPCH
> nation(n_name)* are : *ALGERIA,ARGENTINA,...*
>
> This becomes an issue if the string has ',' (commas), like for example in
> *n_comment* column of *nation* table.
>
> Could someone point out the problem and suggest a solution?
>
> Thank you.
>
> --
> Regards,
> Ashoke
>

--
Regards,
Ashoke