Обсуждение: configure PostgreSQL with PERL: Perl version 5.8 or later is required, but this is .

Andre Mikulec <andre_mikulec@hotmail.com> writes:
> I am getting the error message.

> *** The installed version of Perl, /c/EnterpriseDB/LanguagePack/9.5/x64/Perl-5.20/bin, is too old to use with
PostgreSQL.
> *** Perl version 5.8 or later is required, but this is .
> configure: WARNING:
> *** Without Perl you will not be able to build PostgreSQL from Git.
> *** You can obtain Perl from any CPAN mirror site.
> *** (If you are using the official distribution of PostgreSQL then you do not
> *** need to worry about this, because the Perl output is pre-generated.)
> configure: error: Perl not found

> Why does it say, "Perl version 5.8 or later is required, but this is ."?

It's trying to extract the Perl version number from the output of
"perl -v", and evidently failing to find one.  What does "perl -v"
print with that version of Perl?

            regards, tom lane

On 10/20/2016 10:37 PM, Andre Mikulec wrote:
>
> I m getting this.
>
> TargetUser@TARGETMACH ~
> $ perl -v
> This is perl, v5.8.8 built for msys-64int
>
> 5.8.8 is not right. See later emails.

try  /c/EnterpriseDB/LanguagePack/9.5/x64/Perl-5.20/bin/perl -v

as the perl in your default system path is probably NOT the same as the
one specified in your ./configure building.

my guess is, its barfing on a version number <= 10 due to bad logic...



--
john r pierce, recycling bits in santa cruz

On 10/23/2016 04:05 PM, Andre Mikulec wrote:
>
> I corrected 'PERL=' to include the executable.
> I tried again.
>

Where did you get the Postgres source you are trying to build?




--
Adrian Klaver
adrian.klaver@aklaver.com

On 10/24/2016 11:01 AM, Andre Mikulec wrote:
>
> Never mind.  I figured it out.
>
> Soon I will write up a summary.

What I am not understanding is why you are building from source and not
using the pre-built packages that:

BigSQL:

https://www.bigsql.org/postgresql/installers.jsp

or

EDB:

http://www.enterprisedb.com/products-services-training/pgdownload#windows

provide with their Windows binaries?

>
> Andre Mikulec
> Andre_Mikulec@Hotmail.com
>
>


--
Adrian Klaver
adrian.klaver@aklaver.com

May strait answer is true.

1. I need to change the width of all of my tables because I have very wide data.  There is a compile-time parameter for that.

2. I need to debug, compile and install a modified version of the PostgreSQL extension pl/r.

please don't reply to an existing thread with a completely new topic.     your questions have nothing to do with "Subject: Re: [GENERAL] configure PostgreSQL with PERL: Perl version 5.8 or later is required, but this is ."

-- 
john r pierce, recycling bits in santa cruz

On 10/24/2016 01:19 PM, Andre Mikulec wrote:
>
> May strait answer is true.
>
>
> 1. I need to change the width of all of my tables because I have very
> wide data.  There is a compile-time parameter for that.

Beyond my knowledge, but I understand the custom compile now.

>
> 2. I need to debug, compile and install a modified version of the
> PostgreSQL extension pl/r.
>
>
> Andre Mikulec
> Andre_Mikulec@Hotmail.com
>
>


--
Adrian Klaver
adrian.klaver@aklaver.com

On 10/24/2016 01:49 PM, John R Pierce wrote:
> On 10/24/2016 1:19 PM, Andre Mikulec wrote:
>>
>> May strait answer is true.
>>
>>
>> 1. I need to change the width of all of my tables because I have very
>> wide data.  There is a compile-time parameter for that.
>>
>> 2. I need to debug, compile and install a modified version of the
>> PostgreSQL extension pl/r.
>
> please don't reply to an existing thread with a completely new
> topic.     your questions have nothing to do with *"Subject:* Re:
> [GENERAL] configure PostgreSQL with PERL: Perl version 5.8 or later is
> required, but this is ."

They do if there is no reason to do a custom compile. As it turns out
there was a reason, so we return to regularly scheduled programing:)

>
>
>
>
>
> --
> john r pierce, recycling bits in santa cruz
>


--
Adrian Klaver
adrian.klaver@aklaver.com

Hi Andre,

I've written some C statistics functions for Postgres before, here:

https://github.com/pjungwir/aggs_for_arrays
https://github.com/pjungwir/aggs_for_vecs

They are all really simple, although they operate on arrays, so yours
should be even simpler. The second link has aggregate functions,
including one for sample variance, so that would probably be the easiest
to adapt to your own needs.

In general Postgres C functions use macros to access their arguments
(and to return something). You can read more about it here:

https://www.postgresql.org/docs/current/static/xfunc-c.html

I'm happy to help more if you need it. I've found extending Postgres in
this way to be a lot of fun and very effective at gaining performance.

Good luck!

Paul


On 06/11/2017 09:57 AM, Andre Mikulec wrote:
> SUMMARY
> ------
>
> I am trying to program in PostgreSQL a statistics function.
>
> My old method in SQL is here.
>
> select ( avg(measure) - 0.0 ) / nullif(stddev_pop(case when measure >
> 0.00 then 0.00 else measure end ),0) sortino_true from TABLE/VIEW;
>
> The logic is based on
> SORTINO RATIO: ARE YOU CALCULATING IT WRONG?
> SEPTEMBER 11, 2013
> https://www.rcmalternatives.com/2013/09/sortino-ratio-are-you-calculating-it-wrong/
>
> In PostgreSQL C, how do I get access to the 'X' (data) variable?
>
> DETAILS
> -------
>
> I see a similar function with other variables in here.
> But I am not trying to re-program 'stddev_samp'.
>
> float8_stddev_samp ( lines 2741 through 2767 )
> https://github.com/postgres/postgres/blob/9a34123bc315e55b33038464422ef1cd2b67dab2/src/backend/utils/adt/float.c
>
>   N     = transvalues[0];
>   sumX  = transvalues[1];
>   sumX2 = transvalues[2];
>
> The following article explains it and this makes sense.
>
> Sum of Squares Formula Shortcut
> https://www.thoughtco.com/sum-of-squares-formula-shortcut-3126266
>
> I can match article symbols to PostgreSQL C varaibles.
>
> numerator = N * sumX2 -        sumX * sumX;
>             N * SIGMA(X**2) - (SIGMA(X))**2
>
> BUT I NEED ...
> --------------
>
> However in my case,
> I have this adjustement
>
> "case when  measure > 0.00 then 0.00 else measure end"
>
> So the case seems that I need access to 'X'
> but access to sumX and sumX2 are only shown.
>
> How would I get accress to X?
>
> May anyone point me to a simple good working example?
>
> Thanks,
> Andre Mikulec
> Andre_Mikulec@Hotmail.com
>
>
>