Re: Index AM API changes for deferability

2009/7/15 Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>:
> There is no reason at all to avoid an index AM API change if one is
> useful.

Thinking about this a bit more, perhaps it would be better if I added
an out parameter to the AM for the uniqueness result, rather than
overloading the return value, which is quite ugly:

bool
index_insert(Relation indexRelation,
Datum *values,
bool *isnull,
ItemPointer heap_t_ctid,
Relation heapRelation,
IndexUniqueCheck uniqueness_check,
bool *is_unique);

This would allow me to tidy up some of the code I added to
ExecInsertIndexTuples() which is a bit of a kludge to support
the hash indexes enforcing uniqueness in the future:

http://archives.postgresql.org/pgsql-hackers/2009-07/msg00812.php

Also I could then move the ereport() for unique key violations from
_bt_check_unique() into index_insert() which would allow the
Duplicate key value error patch to be non-btree-specific:

http://archives.postgresql.org/message-id/20090403161658.AFB2.52131E4D@oss.ntt.co.jp
http://archives.postgresql.org/message-id/24655.1238885884@sss.pgh.pa.us

Thoughts?