Re: Refactoring of pg_resetwal/t/001_basic.pl - Mailing list pgsql-hackers

From Peter Eisentraut
Subject Re: Refactoring of pg_resetwal/t/001_basic.pl
Date
Msg-id dc2a324a-8583-4748-9515-fe53553ffe4b@eisentraut.org
Whole thread Raw
In response to Refactoring of pg_resetwal/t/001_basic.pl  (Maxim Orlov <orlovmg@gmail.com>)
Responses Re: Refactoring of pg_resetwal/t/001_basic.pl
List pgsql-hackers
On 21.03.24 17:58, Maxim Orlov wrote:
> In commit 7b5275eec more tests and test coverage were added into 
> pg_resetwal/t/001_basic.pl <http://001_basic.pl>.
> All the added stuff are pretty useful in my view.  Unfortunately, there 
> were some magic constants
> been used.  In overall, this is not a problem.  But while working on 64 
> bit XIDs I've noticed these
> changes and spent some time to figure it out what this magic values are 
> stands fore.
> 
> And it turns out that I’m not the only one.
> 
> So, by Svetlana Derevyanko's suggestion, I made this patch.  I add 
> constants, just like we did
> in verify_heapam tests.

Ok, this sounds like a reasonable idea.

> 
> Sidenote here: in defines in multixact.c TransactionId type used, but 
> I'm sure this is not correct,
> since we're dealing here with MultiXactId and MultiXactOffset.  For now, 
> this is obviously not a
> problem, since sizes of this particular types are equal.  But this will 
> manifest itself when we switch
> to the 64 bits types for MultiXactOffset or MultiXactId.

Please send a separate patch for this if you want to propose any changes.




pgsql-hackers by date:

Previous
From: Peter Geoghegan
Date:
Subject: Re: Optimizing nbtree ScalarArrayOp execution, allowing multi-column ordered scans, skip scan
Next
From: Daniel Gustafsson
Date:
Subject: Re: doc issues in event-trigger-matrix.html