Peter T Mount <peter(at)retep(dot)org(dot)uk> writes:
> Quoting Joe Conway <joseph(dot)conway(at)home(dot)com>:
>> indicating that the practical limit is closer to 1 GB -- not sure why).
> It may be something to do with the 1Gb splitting of the physical files
> representing a table...
No, that's just a coincidence. The reason that TOAST limits fields to
1Gb is that the high-order two bits of the varlena length word were
commandeered as TOAST state indicators. There are now only thirty bits
available to represent the length of a variable-length Datum; hence the
hard limit on field width is 1Gb.
I'd think that the "practical" limit is quite a bit less than that, at
least until we devise an API that lets you read and write toasted values
regards, tom lane
In response to
pgsql-hackers by date
|Next:||From: Hannu Krosing||Date: 2001-08-24 15:33:43|
|Subject: Re: User locks code|
|Previous:||From: Bruce Momjian||Date: 2001-08-24 14:57:28|
|Subject: Re: A couple items on TODO|