| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | Andres Freund <andres(at)anarazel(dot)de> |
| Cc: | Justin Pryzby <pryzby(at)telsasoft(dot)com>, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com>, pgsql-hackers(at)lists(dot)postgresql(dot)org |
| Subject: | Re: Backpatch b61d161c14 (Introduce vacuum errcontext ...) |
| Date: | 2020-06-22 22:15:27 |
| Message-ID: | 2094384.1592864127@sss.pgh.pa.us |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Andres Freund <andres(at)anarazel(dot)de> writes:
> No, I don't think that's a solution. I think it's wrong to have
> something like olderrinfo in the first place. Using a struct with ~25
> members to store the current state of three variables just doesn't make
> sense. Why isn't this just a LVSavedPosition struct or something like
> that?
That seems like rather pointless micro-optimization really; the struct's
not *that* large. But I have a different complaint now that I look at
this code: is it safe at all? I see that the indname field is a pointer
to who-knows-where. If it's possible in the first place for that to
change while this code runs, then what guarantees that we won't be
restoring a dangling pointer to freed memory?
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Tomas Vondra | 2020-06-22 22:25:30 | Re: POC: GROUP BY optimization |
| Previous Message | Thomas Munro | 2020-06-22 21:52:21 | Re: Parallel Seq Scan vs kernel read ahead |