This content has been marked as final.
Show 4 replies
-
1. Re: org.jboss.mq.SpyDestination.writeDest
starksm64 Dec 29, 2005 11:23 AM (in response to arro239)You need to look at how the Serializable contract is used vs this writeDest utility method. They are not the same thing that can simply be encapsulated via an instance specific writeObject.
-
2. Re: org.jboss.mq.SpyDestination.writeDest
arro239 Dec 29, 2005 12:29 PM (in response to arro239)i never meant any specifict exisitng writeObject method should be used instead of exising writeDesc
i mean pieces of code which are related to each specific class inheriting SpyDestination could be moved to the body of those classes - it is discussinble if that would make code clearer - but at least [i hope at least this is true?] that would work faster - we will replace numerous constinues 'instanceof' to one and exactly one call of method -
3. Re: org.jboss.mq.SpyDestination.writeDest
timfox Dec 29, 2005 1:34 PM (in response to arro239)"arro239" wrote:
it is discussinble if that would make code clearer - but at least [i hope at least this is true?] that would work faster - we will replace numerous constinues 'instanceof' to one and exactly one call of method
IMO, the performance gains (if any) of this will be *utterly* insignificant compared to the cost of serialization itself and network io. -
4. Re: org.jboss.mq.SpyDestination.writeDest
arro239 Dec 29, 2005 4:46 PM (in response to arro239)"timfox" wrote:
IMO, the performance gains (if any) of this will be *utterly* insignificant compared to the cost of serialization itself and network io.
that's 105% thue about network IO, and that's true in general [but this IS a part of serialization cost]
my guestion was mostly about if this implementation was chosen for some reason - because for me it's very ugly. so, i was interested if this implementation is good from some point of view - and the non-OOP style is a payment for that