USECS_* constants undefined with float8 timestamps?

On Fri, Jul 29, 2011 at 11:18 AM, Johann 'Myrkraverk' Oskarsson
<johann@2ndquadrant.com> wrote:

Hi all,

I just noticed that the USECS_* constants are not defined when the server
is compiled without integer dates and timestamps.

Explicitly, timestamp.h is

#ifdef HAVE_INT64_TIMESTAMP
#define USECS_PER_DAY   INT64CONST(86400000000)
#define USECS_PER_HOUR  INT64CONST(3600000000)
#define USECS_PER_MINUTE INT64CONST(60000000)
#define USECS_PER_SEC   INT64CONST(1000000)
#endif

Is there a particular reason for this?  Even with float8 timestamps
there are uses for these constants in extensions.

I don't see any particular reason not define them unconditionally.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

Robert Haas wrote:

On Fri, Jul 29, 2011 at 11:18 AM, Johann 'Myrkraverk' Oskarsson
<johann@2ndquadrant.com> wrote:

Hi all,

I just noticed that the USECS_* constants are not defined when the server
is compiled without integer dates and timestamps.

Explicitly, timestamp.h is

#ifdef HAVE_INT64_TIMESTAMP
#define USECS_PER_DAY INT64CONST(86400000000)
#define USECS_PER_HOUR INT64CONST(3600000000)
#define USECS_PER_MINUTE INT64CONST(60000000)
#define USECS_PER_SEC INT64CONST(1000000)
#endif

Is there a particular reason for this? ?Even with float8 timestamps
there are uses for these constants in extensions.

I don't see any particular reason not define them unconditionally.

Well, they are only used by integer dates, so why expand their
visibility? The define does make it clear how they are used. I suppose
if someone wanted to use them outside that case, we could open them up.
It is true that with integer dates now the default, we might find that
someone introduces compile problems by using them outside the integer
dates scope.

--
Bruce Momjian <bruce@momjian.us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ It's impossible for everything to be true. +

Bruce Momjian <bruce@momjian.us> writes:

Robert Haas wrote:

On Fri, Jul 29, 2011 at 11:18 AM, Johann 'Myrkraverk' Oskarsson
<johann@2ndquadrant.com> wrote:

I just noticed that the USECS_* constants are not defined when
the server is compiled without integer dates and timestamps.

[snip]

I don't see any particular reason not define them unconditionally.

Well, they are only used by integer dates, so why expand their
visibility? The define does make it clear how they are used. I
suppose if someone wanted to use them outside that case, we could
open them up. It is true that with integer dates now the default, we
might find that someone introduces compile problems by using them
outside the integer dates scope.

I found a use for them in PL/Java which detects at run-time whether
the server is using floating point or integer dates. The simplest way
was just to use magic numbers instead on the off chance it's compiled
with a server using float dates.

--
Johann Oskarsson http://www.2ndquadrant.com/ |[]
PostgreSQL Development, 24x7 Support, Training and Services --+--
|
Blog: http://my.opera.com/myrkraverk/blog/

Johann 'Myrkraverk' Oskarsson wrote:

Bruce Momjian <bruce@momjian.us> writes:

Robert Haas wrote:

On Fri, Jul 29, 2011 at 11:18 AM, Johann 'Myrkraverk' Oskarsson
<johann@2ndquadrant.com> wrote:

I just noticed that the USECS_* constants are not defined when
the server is compiled without integer dates and timestamps.

[snip]

I don't see any particular reason not define them unconditionally.

Well, they are only used by integer dates, so why expand their
visibility? The define does make it clear how they are used. I
suppose if someone wanted to use them outside that case, we could
open them up. It is true that with integer dates now the default, we
might find that someone introduces compile problems by using them
outside the integer dates scope.

I found a use for them in PL/Java which detects at run-time whether
the server is using floating point or integer dates. The simplest way
was just to use magic numbers instead on the off chance it's compiled
with a server using float dates.

OK, that is a good reason. Done for PG 9.2.

--
Bruce Momjian <bruce@momjian.us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ It's impossible for everything to be true. +