Thread: Re: [COMMITTERS] pgsql: Include the backend ID in the relpath of temporary relations.
Re: [COMMITTERS] pgsql: Include the backend ID in the relpath of temporary relations.
From
Robert Haas
Date:
On Fri, Aug 13, 2010 at 6:38 PM, Robert Haas <robertmhaas@gmail.com> wrote: > On Fri, Aug 13, 2010 at 5:51 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote: >> rhaas@postgresql.org (Robert Haas) writes: >>> Include the backend ID in the relpath of temporary relations. >> >> A couple of the buildfarm members don't like this patch. I think you >> missed making some edits in some dtrace calls. > > Well, I guess that's why we have a buildfarm. Working on it now... I have taken a crack at fixing this but someone who understands DTrace better than I do may want to check and see if the changes look sane. It appears to me that we have no documentation - not even so much as a source code comment - explaining how these probes are supposed to work or what the arguments to each one are intended mean. That may not be ideal. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise Postgres Company
Re: Re: [COMMITTERS] pgsql: Include the backend ID in the relpath of temporary relations.
From
Tom Lane
Date:
Robert Haas <robertmhaas@gmail.com> writes: > I have taken a crack at fixing this but someone who understands DTrace > better than I do may want to check and see if the changes look sane. > It appears to me that we have no documentation - not even so much as a > source code comment - explaining how these probes are supposed to work > or what the arguments to each one are intended mean. http://developer.postgresql.org/pgdocs/postgres/dynamic-trace.html#DTRACE-PROBE-POINT-TABLE (... which you now need to update ...) I think your confusion may stem from the fact that the definition of the buffer-read-done probe was actually wrong, AFAICS. The docs say its last 3 args were bools, which was reasonable, but the definition said int for the first of those. Which is what you want now ... regards, tom lane
Re: Re: [COMMITTERS] pgsql: Include the backend ID in the relpath of temporary relations.
From
Robert Haas
Date:
On Fri, Aug 13, 2010 at 7:23 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote: > Robert Haas <robertmhaas@gmail.com> writes: >> I have taken a crack at fixing this but someone who understands DTrace >> better than I do may want to check and see if the changes look sane. >> It appears to me that we have no documentation - not even so much as a >> source code comment - explaining how these probes are supposed to work >> or what the arguments to each one are intended mean. > > http://developer.postgresql.org/pgdocs/postgres/dynamic-trace.html#DTRACE-PROBE-POINT-TABLE > > (... which you now need to update ...) > > I think your confusion may stem from the fact that the definition of the > buffer-read-done probe was actually wrong, AFAICS. The docs say its > last 3 args were bools, which was reasonable, but the definition said > int for the first of those. Which is what you want now ... No, it was the original patch that mangled that. I think the real problem is that (1) I didn't test with dtrace enabled when writing the patch, or maybe I did somewhere in the middle but not at the end and (2) I didn't realize there were docs. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise Postgres Company