Re: Enums patch v2

Neil Conway wrote:
> On Thu, 2007-02-01 at 22:50 -0500, Bruce Momjian wrote:
>> Where are we on this?
> 
> I can commit to reviewing this. The patch looked fairly solid on a quick
> glance through, but I won't have the cycles to review it properly for a
> week or two.

I've brought this up to date with the operator family stuff now, and the 
attached patch once more applies cleanly against HEAD. Hopefully that'll 
make life a little easier when reviewing it. Thanks.

I got sick of manually shifting the new OID values every time someone 
had added something new to the catalogs, so I moved all of my OIDs up to 
the 9000 range. Attached is a hacky bash script which can move them back 
to something sane. The idea is to run unused_oids first, see where the 
main block of unused OIDs starts, and then run shift_oids on the 
(unzipped) patch file before applying the patch. We discussed something 
similar a while ago, but no-one ever got around to implementing the script.

I've attached the script and left the OIDs in my patch in the upper 
range rather than just running it myself before submitting the patch as 
I reckon that this might be a useful convention for authors of patches 
which add a non-trivial number of OIDs to the catalogs. If there's 
agreement then anyone can feel free to commit the script to CVS or put 
it on the wiki or whatever.

Cheers

Tom
#!/bin/sh

start=$1
end=$2
new_start=$3
filename=$4

if [ -z "$filename" ] ; then
	echo Usage: $0 start-oid end-oid new-start-oid filename
	exit 1
fi

if [ ! -f $filename ] ; then
	echo $0: $filename is not a file
	exit 1
fi

if [ $end -le $start ] ; then
	echo $0: End of OID range must be greater than or equal to the start
	exit 1
fi

start_len=`echo -n $start | wc -c`
end_len=`echo -n $end | wc -c`

if [ $start_len -ne 4 -o $end_len -ne 4 ] ; then
	echo $0: Source OID range must have 4 digits
	exit 1
fi

let new_end=$new_start+$end-$start
if [ $start -le $new_start -a $end -ge $new_start -o $new_start -le $start -a $new_end -ge $start ] ; then
	echo $0: OID ranges may not overlap
	exit 1
fi


i=$start
j=$new_start
while [ $i -le $end ] ; do
	#echo $i $j
	sed -i "s/$i/$j/g" $filename
	let i=i+1
	let j=j+1
done