pypgsql-users

[Pypgsql-users] unQuoteBytea

[Ben L. <pyp...@zu...>]

The pgUnquoteBytea function in libpq is rather slow (it uses scanf). Noticing 
that the pyPgSQL version was much faster I lifted the code and put it into 
libpq.

Do you mind if I submit the change to the libpq developer with full 
acknowledgement that the code came from pyPgSQL?

Thanks,

Ben Lamb.

BTW Unescaping a 200Kb JPEG stored in a PostgreSQL bytea field took 1.5 mins 
before and 0.09 seconds afterwards on a 2Ghz machine!

Re: [Pypgsql-users] unQuoteBytea

[<gh...@gh...>]

Ben Lamb wrote:
> The pgUnquoteBytea function in libpq is rather slow (it uses scanf). Noticing 
> that the pyPgSQL version was much faster I lifted the code and put it into 
> libpq.
> 
> Do you mind if I submit the change to the libpq developer with full 
> acknowledgement that the code came from pyPgSQL?

That's what Open Source is about, right? So go ahead and submit it to 
the PostgreSQL developers at pgs...@po.... Billy wrote 
that particular function if the question of copyright should occur on 
the patches list. pyPgSQL is also licensed under a BSD license so they 
should have no problems accepting the patch.

> BTW Unescaping a 200Kb JPEG stored in a PostgreSQL bytea field took 1.5 mins 
> before and 0.09 seconds afterwards on a 2Ghz machine!

Wow, I wouldn't have imagined that (some platforms') scanf suck so much. 
Doesn't look like libpq's bytea unquoting function was used very much in 
real life, does it?

-- Gerhard

Re: [Pypgsql-users] unQuoteBytea

[Ben L. <pyp...@zu...>]

On Thursday 01 May 2003 1:24 am, you wrote:
> > BTW Unescaping a 200Kb JPEG stored in a PostgreSQL bytea field took 1.5
> > mins before and 0.09 seconds afterwards on a 2Ghz machine!
>
> Wow, I wouldn't have imagined that (some platforms') scanf suck so much.
> Doesn't look like libpq's bytea unquoting function was used very much in
> real life, does it?

I guess not many people use sscanf for parsing large (> 1Kb) amounts of data. 
The function was only introduced in 7.3 so I doubt many people have noticed 
it yet.

I'm using Debian stable on Intel, glibc 2.2.5 with gcc-3.2.2.

Ben.