[COMMITTERS] pgsql: Update comment.

2014-10-22 Thread Heikki Linnakangas
Update comment. The _bt_tuplecompare() function mentioned in comment hasn't existed for a long time. Peter Geoghegan Branch -- master Details --- http://git.postgresql.org/pg/commitdiff/98b37437796bb5fedecf0d0c73cef4aa2ce9d192 Modified Files -- src/backend/utils/sort/tuples

[COMMITTERS] pgsql: Small code cleanup.

2014-10-22 Thread Michael Meskes
Small code cleanup. Declare static variable as static and external as extern. Branch -- master Details --- http://git.postgresql.org/pg/commitdiff/2ae7811db87bf54e8a70aae3cfd9f7cc7eff0bc4 Modified Files -- src/interfaces/ecpg/preproc/ecpg.header |2 +- src/interfaces/

Re: [COMMITTERS] pgsql: Allow empty target list in SELECT.

2014-10-22 Thread Thom Brown
On 15 December 2013 02:23, Tom Lane wrote: > Allow empty target list in SELECT. > > This fixes a problem noted as a followup to bug #8648: if a query has a > semantically-empty target list, e.g. SELECT * FROM zero_column_table, > ruleutils.c will dump it as a syntactically-empty target list, whic

Re: [COMMITTERS] pgsql: Allow empty target list in SELECT.

2014-10-22 Thread Tom Lane
Thom Brown writes: > On 15 December 2013 02:23, Tom Lane wrote: >> Allow empty target list in SELECT. > This commit introduces another bug: > # create table colours (id serial, name text, visible boolean); > CREATE TABLE > # insert into colours (name, visible) values > ('blue',true),('yellow',

Re: [COMMITTERS] pgsql: Allow empty target list in SELECT.

2014-10-22 Thread Thom Brown
On 22 October 2014 21:57, Tom Lane wrote: > Thom Brown writes: > > On 15 December 2013 02:23, Tom Lane wrote: > >> Allow empty target list in SELECT. > > > This commit introduces another bug: > > > # create table colours (id serial, name text, visible boolean); > > CREATE TABLE > > > # insert i

Re: [COMMITTERS] pgsql: Allow empty target list in SELECT.

2014-10-22 Thread Thom Brown
On 22 October 2014 22:06, Thom Brown wrote: > On 22 October 2014 21:57, Tom Lane wrote: > >> Thom Brown writes: >> > On 15 December 2013 02:23, Tom Lane wrote: >> >> Allow empty target list in SELECT. >> >> > This commit introduces another bug: >> >> > # create table colours (id serial, name t

Re: [COMMITTERS] pgsql: Allow empty target list in SELECT.

2014-10-22 Thread Tom Lane
Thom Brown writes: > I can see the same output as you if I get it to ignore my .psqlrc file: > # select * from colours2; > -- > (5 rows) > Turns out it's "\x auto" causing it. Ah. Seems like a bug in the expanded-mode printout logic then. But, in any case, it's been like that a very long time.

[COMMITTERS] pgsql: Ensure libpq reports a suitable error message on unexpected sock

2014-10-22 Thread Tom Lane
Ensure libpq reports a suitable error message on unexpected socket EOF. The EOF-detection logic in pqReadData was a bit confused about who should set up the error message in case the kernel gives us read-ready-but-no-data rather than ECONNRESET or some other explicit error condition. Since the wh

[COMMITTERS] pgsql: Ensure libpq reports a suitable error message on unexpected sock

2014-10-22 Thread Tom Lane
Ensure libpq reports a suitable error message on unexpected socket EOF. The EOF-detection logic in pqReadData was a bit confused about who should set up the error message in case the kernel gives us read-ready-but-no-data rather than ECONNRESET or some other explicit error condition. Since the wh

[COMMITTERS] pgsql: Ensure libpq reports a suitable error message on unexpected sock

2014-10-22 Thread Tom Lane
Ensure libpq reports a suitable error message on unexpected socket EOF. The EOF-detection logic in pqReadData was a bit confused about who should set up the error message in case the kernel gives us read-ready-but-no-data rather than ECONNRESET or some other explicit error condition. Since the wh

[COMMITTERS] pgsql: Ensure libpq reports a suitable error message on unexpected sock

2014-10-22 Thread Tom Lane
Ensure libpq reports a suitable error message on unexpected socket EOF. The EOF-detection logic in pqReadData was a bit confused about who should set up the error message in case the kernel gives us read-ready-but-no-data rather than ECONNRESET or some other explicit error condition. Since the wh

[COMMITTERS] pgsql: Ensure libpq reports a suitable error message on unexpected sock

2014-10-22 Thread Tom Lane
Ensure libpq reports a suitable error message on unexpected socket EOF. The EOF-detection logic in pqReadData was a bit confused about who should set up the error message in case the kernel gives us read-ready-but-no-data rather than ECONNRESET or some other explicit error condition. Since the wh

[COMMITTERS] pgsql: Ensure libpq reports a suitable error message on unexpected sock

2014-10-22 Thread Tom Lane
Ensure libpq reports a suitable error message on unexpected socket EOF. The EOF-detection logic in pqReadData was a bit confused about who should set up the error message in case the kernel gives us read-ready-but-no-data rather than ECONNRESET or some other explicit error condition. Since the wh

[COMMITTERS] pgsql: Minimize calls of pg_class_aclcheck to minimum necessary

2014-10-22 Thread Peter Eisentraut
Minimize calls of pg_class_aclcheck to minimum necessary In a couple of code paths, pg_class_aclcheck is called in succession with multiple different modes set. This patch combines those modes to have a single call of this function and reduce a bit process overhead for permission checking. Autho

[COMMITTERS] pgsql: Add tests for sequence privileges

2014-10-22 Thread Peter Eisentraut
Add tests for sequence privileges Branch -- master Details --- http://git.postgresql.org/pg/commitdiff/a5f7d581943e70235264098c35313d16e932ee0c Modified Files -- src/test/regress/expected/sequence.out | 123 ++ src/test/regress/expected/sequence_