RE: date-time domain overhead in express queries.

Question

DateTimeData objects in Forte are inefficient, both in terms of memory
footprint and instantiation speed. My understanding is that it stores
different parts of the date as separate textdata sub-objects.. Large
footprint also means they take a long time to pass across the network.
We rarely use DataTimeData or any related "domain" type objects. We have
our own date class which stores the entire date and time in a single string
attribute.. Much faster, much smaller footprint, etc. Fetches from the
database use a "to_char(...)" approach to format the date from the database
directly in to the internal string attribute in the correct format, again
for speed..
-Greg
-----Original Message-----
From: Jones, Kerreck [SMTP:[email protected]]
Sent: Tuesday, January 19, 1999 11:36 AM
To: '[email protected]'
Subject: date-time domain overhead in express queries.
In our application, we have an outline window which displays 11 fields, 3
of
which are date time domain.
Our database currently contains about 1500 entries, and the query takes as
long as 70 seconds. If we
remove the DTG domains from the application model, then the query takes
only
35 seconds. Is there any
known problem with the overhead of the express date time domain when doing
queries?
Thanks in advance for any info.
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>-
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>

Answer 1

I have considered this but have not found an elegant solution to the
problem of working accross multiple time-zones.
The application I am working on will ultimately need to work within Europe,
potentially servicing multiple time zones. By converting the date to a
string we lose the time-zone conversion that Forte does when an object
moves partitions and this is likely to cause problems as we then have no
"central" point of application time. If this is not a problem for you then
I would use strings as much as you possibly can.
If anyone has any thougths on how I could use scalar's in this case I'd be
interested to hear them. I've thought about forcing clients to "register"
with my central clock and then doing conversions behind the scenes but it
seems a bit inelegant.
However, one thing cannot be emphasised enough. Using the simple object
types (TextData's, IntegerData's etc) as class attributes is a really bad
idea, and will give you a number of performance penatlies.
Serialisation/De-serialisation of objects is really slow when compared with
scalars.
Scalars take up no 'object space' so you save on garbage collection.
Scalars are tiny when compared with their object equivalents.
Hope that helps.
From: [email protected]
To: [email protected]
Cc: [email protected]
Subject: RE: date-time domain overhead in express queries.
Date: 19 January 1999 18:22
DateTimeData objects in Forte are inefficient, both in terms of memory
footprint and instantiation speed. My understanding is that it stores
different parts of the date as separate textdata sub-objects.. Large
footprint also means they take a long time to pass across the network.
We rarely use DataTimeData or any related "domain" type objects. We have
our own date class which stores the entire date and time in a singlestring
attribute.. Much faster, much smaller footprint, etc. Fetches from the
database use a "to_char(...)" approach to format the date from thedatabase
directly in to the internal string attribute in the correct format, again
for speed..
-Greg
-----Original Message-----
From: Jones, Kerreck [SMTP:[email protected]]
Sent: Tuesday, January 19, 1999 11:36 AM
To: '[email protected]'
Subject: date-time domain overhead in express queries.
In our application, we have an outline window which displays 11 fields,
3
of
which are date time domain.
Our database currently contains about 1500 entries, and the query takesas
long as 70 seconds. If we
remove the DTG domains from the application model, then the query takes
only
35 seconds. Is there any
known problem with the overhead of the express date time domain whendoing
queries?
Thanks in advance for any info.
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive<URL:http://pinehurst.sageit.com/listarchive/>
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>-
To unsubscribe, email '[email protected]' with
'unsubscribe forte-users' as the body of the message.
Searchable thread archive <URL:http://pinehurst.sageit.com/listarchive/>