2010/12/29 Tom Lane
<tgl@sss.pgh.pa.us>I think they're fundamentally different things, because the previously
proposed patch is an extension of the machine-readable archive format,
and has to remain so because of the expectation that people will want
to use parallel restore with it. Joel is arguing for a split-up of
the text dump format.
Yes, exactly.
My patch is of course also a lot smaller :-)
pg_dump-directory.diff.........................: 112 853 bytes
pg-dump-split-plain-text-files-9.1devel.patch..: 5 579 bytes
I just tried the pg_dump-directory.diff patch.
The only thing is has in common with my patch is it writes data to different files, and it's only the data which is splitted into different files, the schema appears to go into the single file "TOC".
Example, pg_dump-directory.diff:
$ ./pg_dump -f /crypt/dirpatch -F d -s glue
$ ls -la /crypt/dirpatch/
TOC
(1 file)
$ rm -rf /crypt/dirpatch
$ ./pg_dump -f /crypt/dirpatch -F d glue
$ ls /crypt/dirpatch/
6503.dat
6504.dat
...lots of files...
6871.dat
6872.dat
6873.dat
6874.dat
TOC
Example, pg_dump --split patch:
$ pg_dump -f /crypt/splitpatch -F p --split -s glue
$ ls /crypt/splitpatch*
/crypt/splitpatch (file)
/crypt/splitpatch-split: (directory)
myschema1
myschema2
public
$ ls /crypt/splitpatch-split/public/
AGGREGATE
CONSTRAINT
FK_CONSTRAINT
FUNCTION
INDEX
SEQUENCE
TABLE
TRIGGER
TYPE
VIEW
$ ls /crypt/splitpatch-split/public/FUNCTION/
myfunc.sql
otherfunc.sql
$ cat /crypt/splitpatch
--
-- PostgreSQL database dump
--
SET statement_timeout = 0;
SET client_encoding = 'UTF8';
SET standard_conforming_strings = off;
SET check_function_bodies = false;
SET client_min_messages = warning;
SET escape_string_warning = off;
...etc...
\i /crypt/splitpatch-split/public/FUNCTION/myfunc.sql
\i /crypt/splitpatch-split/public/FUNCTION/otherfunc.sql
--
Best regards,
Joel Jacobson
Glue Finance
