Re: Pointers across Partitions...For YourInformation

Question

At 09:41 AM 12/3/97 -0500, Sivaram S Ghorakavi wrote:
Hello Folks,
This is just FYI.
My application has got a Server Partition, with a Service Object as aninteraction to a wrapper in 'C' for a DLL, and a Client Partition, which
basically access the SO on each user interaction. The most of the functions
in 'C' code expect "pointer to char/pointer to structure etc.,". My
application CRASHES by declaring the pointers and allocating the memory and
passing them from client to server partition. But if I pass the forte data
values, like TextData etc., to the SO and let that do the CASTING of the
values and passing them to the methods on the Server Partion makes my
application HAPPY.
>
Rule of Thumb: Never pass the pointers across partitions. Pass the fortedatavalues and let the other partitionon handle them.
>
Regards,
Sivaram S Ghorakavi mailto:[email protected]
International Business Corporation http://www.ibcweb.com/
Actually, this makes a lot of sense when you think about it.
First, a little about pointers. As most of you probably know, a pointer is
simply a variable (usually a long, in C++) that holds a memory address
where the actual data that you are interested resides. In other words, the
pointer "points" to the memory location of the data.
Imagine an application running on some networked computer (call it "Machine
A"). This application has a partition (call it "Partition 1") that is
using some data stored in memory location 123456. In Partition 1, a
pointer variable ("P") has the value "123456", thus P "points" to the data.
Now lets pass P to another partition ("Partition 2") running on another
computer (call this "Machine B"). Remember, P really only contains the
value "123456". When it is accessed by the code in Partition 2, the
application would actually be looking for data in memory location 123456
***on Machine B***! Clearly, this is a violation and would cause all kinds
of problems.
In this respect, Forte doesn't differentiate between partitions running on
different computers vs. partitions running on the same computer, so it
raises an exception whenever a program tries to pass pointers across any
partition boundries.
Passing objects is different, however. Most of you know that Forte uses
what is called "copy semantics" when passing objects across partition
boundries. This means that, while to the developer it may appear that an
object received from a remote partition is the "same" object as the one
that was passed by that remote partition, it is actually a copy - a
physically different object containing all of the same state information as
the original object. Thus, the "address" of an object doesn't really come
into play across partition boundries.
I'm not going to take the time to explain here how anchored objects work,
but they are really object-based, not pointer-based, as well.
Viola! Mr. Ghorakavi's rule of thumb explained!
James Urquhart [email protected]
Product Manager phone: [phone]
Forte Software, Inc. fax: [phone]

Answer 1

Maybe, just maybe, you can use gparted to move the excess space in your new partition which is above your existing partition size after first identifying it as swap.  Then move it to the end of the drive, leaving space for a new partition which is the same size as your existing partition.
That, maybe, will accept the dd transfer since the partition size is identical.  After all is done, delete the swap.
I hope I made sense.....