[kronolith] Blank Kronolith page

[Jan Schneider]

Zitat von Dan Wilson <dan@acucore.com>:

> : > OK... so I finally decided to give up on mcal and go with sql as
> well...
> : > but
> : > I'm running PostgreSQL instead of MySQL and I get this error:
> : >
> : > ERROR: parser: parse error at or near "{"
> : >
> : > Of course, that's just part of the error...   it's not liking the
> time
> : > stamp
> : > syntax used... here's the full sql statement that it doesn't like:
> : >
> : > SELECT DISTINCT e.event_id, e.event_recurType
> : > FROM kronolith_events e
> : > WHERE
> : >   e.calendar_id = 'dan@acucore.com'
> : >   AND ((e.event_start < {ts '2001-11-18 00:00:00.000'} AND
> e.event_end >
> : >
> : > {ts '2001-11-17 00:00:00.000'}) OR (e.event_recurEnddate >= {ts
> : > '2001-11-17
> : > 00:00:00.000'} AND e.event_start <= {ts '2001-11-18 00:00:00.000'}
> AND
> : > e.event_recurType != 0))
> : >
> : > Postgres' timestamps use different syntax, and I am too unfamiliar
> with
> : > PEAR to
> : > know how to fix it and keep it compatible with other db's.
> :
> : I know Chuck worked hard to find a syntax as compatible as possible, so
> if
> : anyone knows a better one...
> :
> : Jan.
> 
> Postgres could handle this without anything around the timestamp
> string.
> Just '2001-11-18 00:00:00.000' would work great.  I also tested this
> with
> MySQL, and it accepted the query, but I don't know if the calculations
> would
> be correct.  I'm assuming that the {ts } syntax is the way to cast text
> to a
> timestamp for MySQL?  In Postgres, usually the typecasting is done
> automatically, but in the event that it requires a explicit cast, the
> syntax
> would be '2001-11-18 00:00:00.000'::timestamp.
> 
> Could someone test to see if we can strip out the {ts ...} syntax and
> the
> calculations still work for mysql?

MySQL has no problems with this syntax but I rather see MSSQL a little 
problematic.

Chuck?
 
> I also had to change the create table statement in kronolith.sql.  The
> default of 0's on the timestamp doesn't work for postgres.  It requires
> a
> valid date.  I changed it to "now()" which is the time of record
> insertion.

That's not so good. Does PostgreSQL generate a default value automatically 
if you add a timestamp field to the table structure NOT NULL without 
specifying a default?

Jan.

:::::::::::::::::::::::::::::::::::::::: 
AMMMa AG - discover your knowledge
:::::::::::::::::::::::::::
Detmolder Str. 25-33 :: D-33604 Bielefeld
fon +49.521.96878-0 :: fax  +49.521.96878-20
http://www.ammma.de
::::::::::::::::::::::::::::::::::::::::::::::