to see which ones do not appear. You can also use

the <filename>duplicate_oids</filename> script to check for mistakes.

(<filename>genbki.pl</filename> will assign OIDs for any rows that

didn't get one hand-assigned to them and also detect duplicate OIDs

at compile time.)

didn't get one hand-assigned to them, and it will also detect duplicate

OIDs at compile time.)

</para>

<para>

When choosing OIDs for a patch that is not expected to be committed

immediately, best practice is to use a group of more-or-less

consecutive OIDs starting with some random choice in the range

8000&mdash;9999. This minimizes the risk of OID collisions with other

patches being developed concurrently. To keep the 8000&mdash;9999

range free for development purposes, after a patch has been committed

to the master git repository its OIDs should be renumbered into

available space below that range. Typically, this will be done

near the end of each development cycle, moving all OIDs consumed by

patches committed in that cycle at the same time. The script

<filename>renumber_oids.pl</filename> can be used for this purpose.

If an uncommitted patch is found to have OID conflicts with some

recently-committed patch, <filename>renumber_oids.pl</filename> may

also be useful for recovering from that situation.

</para>

<para>

Because of this convention of possibly renumbering OIDs assigned by

patches, the OIDs assigned by a patch should not be considered stable

until the patch has been included in an official release. We do not

change manually-assigned object OIDs once released, however, as that

would create assorted compatibility problems.

</para>

<para>

The OID counter starts at 10000 at the beginning of a bootstrap run.

If a row from a source other than <filename>postgres.bki</filename>

is inserted into a table that requires OIDs, then it will receive an

OID of 10000 or above.

OID of 10000 or above. For example, objects created while running

the <filename>information_schema.sql</filename> script receive such

OIDs.

</para>

<para>

OIDs assigned during normal database operation are constrained to be

16384 or higher. This leaves the range 10000&mdash;16383 available

for OIDs assigned automatically during bootstrap. These OIDs are not

considered stable, and may change from one installation to another.

</para>

</sect2>

}

if (/^DECLARE_TOAST\(\s*(\w+),\s*(\d+),\s*(\d+)\)/)

{

push @{$catalog{toasting} },

use strict;

use warnings;

use FindBin;

use Getopt::Long;

chdir$FindBin::RealBinordie"could not cd to$FindBin::RealBin:$!\n";

use lib"$FindBin::RealBin/../../backend/catalog/";

use Catalog;

my$FirstGenbkiObjectId =

Catalog::FindDefinedSymbol('access/transam.h','..','FirstGenbkiObjectId');

my$output_path ='';

my$first_mapped_oid = 0;

my$last_mapped_oid =$FirstGenbkiObjectId - 1;

my$target_oid = 0;

GetOptions(

'output=s'=> \$output_path,

'first-mapped-oid=i'=> \$first_mapped_oid,

'last-mapped-oid=i'=> \$last_mapped_oid,

'target-oid=i'=> \$target_oid) || usage();

die"Unexpected non-switch arguments.\n"if@ARGV;

if$first_mapped_oid <= 0;

if$last_mapped_oid <$first_mapped_oid;

if$target_oid <= 0;

if$target_oid >=$first_mapped_oid &&$target_oid <=$last_mapped_oid;

if ($output_pathne'' &&substr($output_path, -1)ne'/')

{

$output_path .='/';

}

my@header_files = (glob("pg_*.h"),qw(indexing.h toasting.h));

my$oids = Catalog::FindAllOidsFromHeaders(@header_files);

my%oidhash;

@oidhash{@$oids} =undef;

my%maphash;

my$next_oid =$target_oid;

for (

my$mapped_oid =$first_mapped_oid;

$mapped_oid <=$last_mapped_oid;

$mapped_oid++)

{

nextif !exists$oidhash{$mapped_oid};

$next_oid++

while (

exists$oidhash{$next_oid}

|| ($next_oid >=$first_mapped_oid

&&$next_oid <=$last_mapped_oid));

if$next_oid >=$FirstGenbkiObjectId;

$maphash{$mapped_oid} =$next_oid;

$next_oid++;

}

die"There are no OIDs in the mapped range.\n"if$next_oid ==$target_oid;

foreachmy$input_file (@header_files)

{

$input_file =~/(\w+)\.h$/

ordie"Input file$input_file needs to be a .h file.\n";

my$catname =$1;

nextif$catname =~/_d$/;

open(my$ifd,'<',$input_file) ||die"$input_file:$!";

my$output_file ="$output_path$catname.h";

my$tmp_output_file ="$output_file.tmp";

openmy$ofd,'>',$tmp_output_file

ordie"can't open$tmp_output_file:$!";

my$changed = 0;

while (<$ifd>)

{

my$line =$_;

if ($line =~m/^(DECLARE_TOAST\(\s*\w+,\s*)(\d+)(,\s*)(\d+)\)/)

{

my$oid2 =$2;

my$oid4 =$4;

if (exists$maphash{$oid2})

{

$oid2 =$maphash{$oid2};

my$repl =$1 .$oid2 .$3 .$oid4 .")";

$line =~s/^DECLARE_TOAST\(\s*\w+,\s*\d+,\s*\d+\)/$repl/;

$changed = 1;

}

if (exists$maphash{$oid4})

{

$oid4 =$maphash{$oid4};

my$repl =$1 .$oid2 .$3 .$oid4 .")";

$line =~s/^DECLARE_TOAST\(\s*\w+,\s*\d+,\s*\d+\)/$repl/;

$changed = 1;

}

}

elsif (

$line =~m/^(DECLARE_(UNIQUE_)?INDEX\(\s*\w+,\s*)(\d+)(,\s*.+)\)/)

{

if (exists$maphash{$3})

{

my$repl =$1 .$maphash{$3} .$4 .")";

$line =~

s/^DECLARE_(UNIQUE_)?INDEX\(\s*\w+,\s*\d+,\s*.+\)/$repl/;

$changed = 1;

}

}

elsif ($line =~m/^CATALOG\((\w+),(\d+),(\w+)\)/)

{

if (exists$maphash{$2})

{

my$repl =

"CATALOG(" .$1 ."," .$maphash{$2} ."," .$3 .")";

$line =~s/^CATALOG\(\w+,\d+,\w+\)/$repl/;

$changed = 1;

}

if ($line =~m/BKI_ROWTYPE_OID\((\d+),(\w+)\)/)

{

if (exists$maphash{$1})

{

my$repl =

"BKI_ROWTYPE_OID(" .$maphash{$1} ."," .$2 .")";

$line =~s/BKI_ROWTYPE_OID\(\d+,\w+\)/$repl/;

$changed = 1;

}

}

}

elsif ($line =~m/^(\s*#\s*define\s+\w+\s+)(\d+)\b/)

{

if (($catnameeq'indexing' ||$catnameeq'toasting')

&&exists$maphash{$2})

{

my$repl =$1 .$maphash{$2};

$line =~s/^\s*#\s*define\s+\w+\s+\d+\b/$repl/;

$changed = 1;

}

}

print$ofd$line;

}

close$ifd;

close$ofd;

if ($changed ||$output_pathne'')

{

rename$tmp_output_file,$output_file

ordie"can't rename$tmp_output_file to$output_file:$!";

}

{

ordie"can't unlink$tmp_output_file:$!";

}

}

foreachmy$input_file (glob("pg_*.dat"))

{

$input_file =~/(\w+)\.dat$/

ordie"Input file$input_file needs to be a .dat file.\n";

my$catname =$1;

open(my$ifd,'<',$input_file) ||die"$input_file:$!";

my$output_file ="$output_path$catname.dat";

my$tmp_output_file ="$output_file.tmp";

openmy$ofd,'>',$tmp_output_file

ordie"can't open$tmp_output_file:$!";

my$changed = 0;

while (<$ifd>)

{

my$line =$_;

if ($line =~m/\b(oid\s*=>\s*)'(\d+)'/)

{

if (exists$maphash{$2})

{

my$repl =$1 ."'" .$maphash{$2} ."'";

$line =~s/\boid\s*=>\s*'\d+'/$repl/;

$changed = 1;

}

}

if ($line =~m/\b(array_type_oid\s*=>\s*)'(\d+)'/)

{

if (exists$maphash{$2})

{

my$repl =$1 ."'" .$maphash{$2} ."'";

$line =~s/\barray_type_oid\s*=>\s*'\d+'/$repl/;

$changed = 1;

}

}

print$ofd$line;

}

close$ifd;

close$ofd;

if ($changed ||$output_pathne'')

{

rename$tmp_output_file,$output_file

ordie"can't rename$tmp_output_file to$output_file:$!";

}

{

ordie"can't unlink$tmp_output_file:$!";

}

}

{

my$last =$FirstGenbkiObjectId - 1;

die<<EOM;

}

o doc/src/sgml/ref manual pages

* Ports

o update config.guess and config.sub at the start of beta

(from http://savannah.gnu.org/projects/config)

o update ports list in doc/src/sgml/installation.sgml

o update platform-specific FAQ's, if needed

Pre-Beta Tasks

==============

These things should be done at least once per development cycle.

Typically we do them between feature freeze and start of beta test,

but there may be reasons to do them at other times as well.

* Renumber any manually-assigned OIDs between 8000 and 9999

to lower numbers, using renumber_oids.pl (see notes in bki.sgml)

* Update config.guess and config.sub

(from http://savannah.gnu.org/projects/config)

* Update inet/cidr data types with newest Bind patches