Re: BUG #1502: hash_seq_search might return removed entry

Tom Lane wrote:

>Thomas Hallgren  writes:
>  
>
>>Would such a patch be accepted?
>>    
>>
>
>Seems like a brute-force solution.  I'd look first at whether
>AtCommit_Portals could just restart its hashtable scan after
>each deletion; if that seems too inefficient, modify the hash
>table entries themselves to carry a "portal already deleted"
>flag.
>  
>
Yes, a static flag indicating that a deletion has occured will work fine 
since all portals that has not been perused but not dropped now has an 
InvalidSubTransactionId. I'll do it that way then.

Regards,
Thomas Hallgren