Re: Backpatch b61d161c14 (Introduce vacuum errcontext ...) - Mailing list pgsql-hackers

From Justin Pryzby
Subject Re: Backpatch b61d161c14 (Introduce vacuum errcontext ...)
Date
Msg-id 20200626015517.GW4107@telsasoft.com
Whole thread Raw
In response to Re: Backpatch b61d161c14 (Introduce vacuum errcontext ...)  (Amit Kapila <amit.kapila16@gmail.com>)
Responses Re: Backpatch b61d161c14 (Introduce vacuum errcontext ...)
List pgsql-hackers
On Thu, Jun 25, 2020 at 02:31:51PM +0530, Amit Kapila wrote:
> I have looked at both the patches (using separate variables (by
> Justin) and using a struct (by Andres)) and found the second one bit
> better.

Thanks for looking.

> I have improved some comments in the code and for now, kept as two
> patches (a) one for improving the error info for index (mostly
> Justin's patch based on Tom's idea) and (b) the other to generally
> improve the code in this area (mostly Andres's patch).

And thanks for separate patchen :)

> I have done some testing with both the patches and would like to do
> more unless there are objections with these.

Comments:

>         * The index name is saved only during this phase and restored immediately

=> I wouldn't say "only" since it's saved during lazy_vacuum: index AND cleanup.

>update_vacuum_error_info(LVRelStats *errinfo, LVSavedErrInfo *oldpos, int phase,

=> You called your struct "LVSavedErrInfo" but the variables are still called
"pos".  I would call it olderrinfo or just old.

Also, this doesn't (re)rename the "cbarg" stuff that Alvaro didn't like, which
was my 0001 patch.

-- 
Justin



pgsql-hackers by date:

Previous
From: David Rowley
Date:
Subject: Re: Keep elog(ERROR) and ereport(ERROR) calls in the cold path
Next
From: Amit Kapila
Date:
Subject: Re: Backpatch b61d161c14 (Introduce vacuum errcontext ...)