Regression testing for psql
in [PgSql]
|
Prev: [HACKERS] Clearing psql's input buffer after auto-reconnect
Next: (9.1) btree_gist support for searching on "notequals"
From: Stephen Frost on 24 May 2010 22:51 * Stephen Frost (sfrost(a)snowman.net) wrote: > Add regression testing for psql backslash commands > > This patch adds rather extensive regression testing > of the psql backslash commands. Hopefully this will > minimize issues such as the one which cropped up > recently with \h segfaulting. Note that we don't > currently explicit check all the \h options and that > pretty much any catalog changes will mean that this > needs to also be updated. Still, it's a start, we can > reduce the set of tests if that makes sense or they > become a problem. And.. it's way too big to send to the list. The patch is available here: http://snowman.net/~sfrost/psql-regress-help.patch Of course, if people want to suggest tests that just shouldn't be included, I can go through and strip things out. Thanks, Stephen
From: Robert Haas on 24 May 2010 23:38 On Mon, May 24, 2010 at 10:51 PM, Stephen Frost <sfrost(a)snowman.net> wrote: > * Stephen Frost (sfrost(a)snowman.net) wrote: >> Add regression testing for psql backslash commands >> >> This patch adds rather extensive regression testing >> of the psql backslash commands. Hopefully this will >> minimize issues such as the one which cropped up >> recently with \h segfaulting. Note that we don't >> currently explicit check all the \h options and that >> pretty much any catalog changes will mean that this >> needs to also be updated. Still, it's a start, we can >> reduce the set of tests if that makes sense or they >> become a problem. > > And.. it's way too big to send to the list. The patch is available > here: > > http://snowman.net/~sfrost/psql-regress-help.patch > > Of course, if people want to suggest tests that just shouldn't be > included, I can go through and strip things out. Well... I'm a little reluctant to believe that we should have 3.3M of tests for the entire backend and 5M of tests just for psql. Then, too, there's the fact that many of these tests fail on my machine because my username is not sfrost, and/or because of row-ordering differences on backslash commands without enough ORDER BY to fully determine the output order. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise Postgres Company -- Sent via pgsql-hackers mailing list (pgsql-hackers(a)postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
From: Stephen Frost on 25 May 2010 06:23 * Robert Haas (robertmhaas(a)gmail.com) wrote: > > Of course, if people want to suggest tests that just shouldn't be > > included, I can go through and strip things out. > > Well... I'm a little reluctant to believe that we should have 3.3M of > tests for the entire backend and 5M of tests just for psql. Then, > too, there's the fact that many of these tests fail on my machine > because my username is not sfrost, and/or because of row-ordering > differences on backslash commands without enough ORDER BY to fully > determine the output order. Yeah, you know, I had fully intended to go grepping through the output last night to check for things like that, but my wife decided I needed sleep instead. :) Sorry about that. Still, it's more of a general proposal than something I think should be committed as-is. Should we try to deal with those kinds of differences, or just eliminate the tests which are dependent on username, etc? It definitely strikes me that there's a fair bit of code in psql we're not exercising in some fashion in the regression suite... :/ Thanks, Stephen
From: Peter Eisentraut on 26 May 2010 06:01 On tis, 2010-05-25 at 06:23 -0400, Stephen Frost wrote: > * Robert Haas (robertmhaas(a)gmail.com) wrote: > > > Of course, if people want to suggest tests that just shouldn't be > > > included, I can go through and strip things out. > > > > Well... I'm a little reluctant to believe that we should have 3.3M of > > tests for the entire backend and 5M of tests just for psql. Then, > > too, there's the fact that many of these tests fail on my machine > > because my username is not sfrost, and/or because of row-ordering > > differences on backslash commands without enough ORDER BY to fully > > determine the output order. > > Yeah, you know, I had fully intended to go grepping through the output > last night to check for things like that, but my wife decided I needed > sleep instead. :) Sorry about that. Still, it's more of a general > proposal than something I think should be committed as-is. Should we > try to deal with those kinds of differences, or just eliminate the tests > which are dependent on username, etc? It definitely strikes me that > there's a fair bit of code in psql we're not exercising in some fashion > in the regression suite... :/ Maybe pg_regress is not the right framework to test that sort of thing. -- Sent via pgsql-hackers mailing list (pgsql-hackers(a)postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
From: Stephen Frost on 26 May 2010 11:09
* Peter Eisentraut (peter_e(a)gmx.net) wrote: > Maybe pg_regress is not the right framework to test that sort of thing. Perhaps, but if not, then what? And how can we avoid writing a bunch of new code that would then need to be checked itself..? Thanks, Stephen |