> my questions are: > 1. is it a bug and will it be fixed? > 2. if it is a bug - is it in hstore? plperl? my code? My guess would be that plperl isn't freeing it's result set storage until the end of the transaction. Might not be classed as a bug, but certainly an inefficiency.
not good - but - if it is pl/perl only issue - i can live with it.
> 3. i can do the migration using small parts - let's say 100 records at a > time, disconnect, reconnect, convert next 100 records. but - will i be safe > later on during standard work? Once the connection is closed, all memory should be freed. But, it looks to me like you might be able to replace the plperl function by just a straight query. That should be faster too.
really? i was thinking really hard on how to do it in sql, but didn't found any way to achieve it. actually - i dont really think it would be possible at all in standard sql. but then - maybe i'm wrong.
