Thread: Custom datestyle for timestamps
Hi, SET DATESTYLE takes predefined keywords such as ISO or US as arguments, but I can't find a way to specify a custom format string for timestamps. What I'd like to find is an equivalent to Oracle's ALTER SESSION SET nls_date_format='DD/MM/YYYY HH24' for example, where the format follows the sames rules than to_char and to_date. This sets an implicit format for every subsequent text<->date conversion. Is there a way I could achieve that with postgres? A psql-only solution would be good enough for me. Thanks, -- Daniel
On Thursday 05 March 2009 00:19:02 Daniel Verite wrote: > SET DATESTYLE takes predefined keywords such as ISO or US as arguments, > but I can't find a way to specify a custom format string for > timestamps. There is no support for that. > What I'd like to find is an equivalent to Oracle's > ALTER SESSION SET nls_date_format='DD/MM/YYYY HH24' for example, > where the format follows the sames rules than to_char and to_date. > This sets an implicit format for every subsequent text<->date > conversion. ... or that.
Peter Eisentraut wrote: > On Thursday 05 March 2009 00:19:02 Daniel Verite wrote: > > SET DATESTYLE takes predefined keywords such as ISO or US as arguments, > > but I can't find a way to specify a custom format string for > > timestamps. > > There is no support for that. > > > What I'd like to find is an equivalent to Oracle's > > ALTER SESSION SET nls_date_format='DD/MM/YYYY HH24' for example, > > where the format follows the sames rules than to_char and to_date. > > This sets an implicit format for every subsequent text<->date > > conversion. > > ... or that. I think it would valuable to have at least a default format that doesn't include the sub-second precision, since it typically adds 7 characters that enlarge output columns in a way that is generally useless to the human eyes. Do other users agree with that? Best regards, -- Daniel
"Daniel Verite" <daniel@manitou-mail.org> writes: > I think it would valuable to have at least a default format that > doesn't include the sub-second precision, since it typically adds 7 > characters that enlarge output columns in a way that is generally > useless to the human eyes. Do other users agree with that? If you think it's useless, just store your timestamps in a timestamp(0) column. regards, tom lane