| Lists: | pgsql-hackers |
|---|
| From: | Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> |
|---|---|
| To: | Pg Hackers <pgsql-hackers(at)postgresql(dot)org> |
| Cc: | Noah Misch <noah(at)leadboat(dot)com>, Kevin Grittner <kevin(dot)grittner(at)wicourts(dot)gov>, Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com> |
| Subject: | isolationtester and invalid permutations |
| Date: | 2011-10-26 01:13:24 |
| Message-ID: | 1319591143-sup-7036@alvh.no-ip.org |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Lists: | pgsql-hackers |
Instead of simply aborting a spec that specifies running commands on
blocked sessions (what we call an invalid permutation), it seems more
useful to report the problem, cleanup the sessions, and continue with
the next permutation.
This, in conjunction with the dry-run patch I submitted earlier, makes
it easier to determine a working spec: dry-run the spec; copy the
so-generated permutation lines into the spec; run the spec normally,
which reports the invalid permutations; comment out the invalid
permutations from the spec; done.
The attached patch, again from Alexander Shulgin (with some tweaks from
me) does that.
Comments?
--
Álvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>
| Attachment | Content-Type | Size |
|---|---|---|
| isotester-allow-invalid-perms.patch | application/octet-stream | 3.6 KB |
| From: | Robert Haas <robertmhaas(at)gmail(dot)com> |
|---|---|
| To: | Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> |
| Cc: | Pg Hackers <pgsql-hackers(at)postgresql(dot)org>, Noah Misch <noah(at)leadboat(dot)com>, Kevin Grittner <kevin(dot)grittner(at)wicourts(dot)gov>, Heikki Linnakangas <heikki(dot)linnakangas(at)enterprisedb(dot)com> |
| Subject: | Re: isolationtester and invalid permutations |
| Date: | 2011-10-27 12:09:19 |
| Message-ID: | CA+Tgmob4vFWJZHiwZ=QtG3EaV_AfBN5DHM8n57LVD70DzA=D7g@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Lists: | pgsql-hackers |
On Tue, Oct 25, 2011 at 9:13 PM, Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> wrote:
> Instead of simply aborting a spec that specifies running commands on
> blocked sessions (what we call an invalid permutation), it seems more
> useful to report the problem, cleanup the sessions, and continue with
> the next permutation.
>
> This, in conjunction with the dry-run patch I submitted earlier, makes
> it easier to determine a working spec: dry-run the spec; copy the
> so-generated permutation lines into the spec; run the spec normally,
> which reports the invalid permutations; comment out the invalid
> permutations from the spec; done.
>
> The attached patch, again from Alexander Shulgin (with some tweaks from
> me) does that.
>
> Comments?
Seems sensible. I think we should avoid including invalid
permutations in our regression test suite, but this still seems useful
for the reasons you mention.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company