static variables?

  • Thread starter Thread starter yellow1912
  • Start date Start date
Y

yellow1912

Hi, I'm very new to C#, in fact I dont know anything about it. Im
having problem with its static variables. I used static variables alot
in C++, but seem like C# doesnt allow it inside a function the way C#
do, so how can you handle recursive functions that need static
variable.

One more thing, I have a variable like this:
OracleDataReader dr = GetDataReader(sql,conn, out errMsg);
How can I make dr static to use in my recursive function?

PS: Sorry to sound like a newbie, but Im actually am. I'd really
appreciate any help. Thanks
 
[email protected] said:
Hi, I'm very new to C#, in fact I dont know anything about it. Im
having problem with its static variables. I used static variables alot
in C++, but seem like C# doesnt allow it inside a function the way C#
do, so how can you handle recursive functions that need static
variable.
Click to expand...
You either pass the variable via a parameter or you use a local class field.
One more thing, I have a variable like this:
OracleDataReader dr = GetDataReader(sql,conn, out errMsg);
How can I make dr static to use in my recursive function?
Click to expand...

public class C
{
....
OracleDataReader dr;
public void SetConnection()
{
dr = GetDataReader(sql, conn,out errMsg);
}
}


I'd strongly consider passing the reader to the next method as a parameter
however.
 
Yrsh, it kind of sucks that you can't declare static method variables, but
there you are. Declare them at class scope, as private fields, instead.

--
HTH,

Kevin Spencer
Microsoft MVP
..Net Developer
There's a seeker born every minute.
- Dr. "Happy" Harry Cox
 
Thanks^^. But then wouldnt the dr be cleared each time the function is
called?

Here is my code, and I know it's wrong obviously, but cant figure out
how to fix it
private static bool Connection( string fromloc, string
toloc,OracleConnection conn)
{

string sql = "SELECT FLIGHTNO,LEGORDER,TOLOC FROM LEG WHERE FROMLOC =
'" + fromloc + "'";
string errMsg = "";

OracleDataReader dr = GetDataReader( sql, conn, out errMsg );
if ( errMsg != "" )
{
Console.WriteLine( errMsg );
return false;
}
if ( dr == null )
{
Console.WriteLine( "No result" );
return false;
}
while( dr.Read() )
{

if (dr[2].ToString()==toloc)
{
Console.WriteLine("flight: " + dr[0] + " legorder: " + dr[1]);
return true;
}
else
return(Connection(dr[2].ToString(),toloc,conn));
}


//dr.Close(); //close the DataReader
//dr.Dispose();
//conn.Close(
 
I do like static method variables but it appears the CLR does not
support it. VB.NET supports static method variables - its compiler
compiles such variables into private class fields.
 
But then wouldnt the dr be cleared each time the function is
called?
Click to expand...
Then you could initialize the var at declaration or in constructor.
If that is not posible, in the method check and only init if it is null.
 
Im trying to understand it.... Can anyone give me an example (any) or
better on the code I posted:

Task: given start-end location, find all possible connections.
If the function runs the first time and I get the first dr, then it
calls itself and i get the 2nd dr,... so on till the n dr. Then it goes
back from n to n-1.... to 1, how can I keep every dr from being deleted
each time the function called itself? Or is there a better way to solve
this prblem?

private static bool Connection( string fromloc, string
toloc,OracleConnection conn)
{

string sql = "SELECT FLIGHTNO,LEGORDER,TOLOC FROM LEG WHERE FROMLOC =
'" + fromloc + "'";
string errMsg = "";

OracleDataReader dr = GetDataReader( sql, conn, out errMsg );
if ( errMsg != "" )
{
Console.WriteLine( errMsg );
return false;
}

if ( dr == null )
{
Console.WriteLine( "No result" );
return false;
}

while( dr.Read() )
{

if (dr[2].ToString()==toloc)
{
Console.WriteLine("flight: " + dr[0] + " legorder: " + dr[1]);
return true;
}

else
return(Connection(dr[2].ToString(),toloc,conn));

}
}
 
[email protected] said:
Im trying to understand it.... Can anyone give me an example (any) or
better on the code I posted:

Task: given start-end location, find all possible connections.
If the function runs the first time and I get the first dr, then it
calls itself and i get the 2nd dr,... so on till the n dr. Then it goes
back from n to n-1.... to 1, how can I keep every dr from being deleted
each time the function called itself? Or is there a better way to solve
this prblem?
Click to expand...

A local field would work.

Frankly, I don't see where you are going with your code here, why do you
need to keep old DataReaders around?
 
Kevin Spencer said:
Yrsh, it kind of sucks that you can't declare static method variables, but
there you are. Declare them at class scope, as private fields, instead.
Click to expand...

Static variables are little more than fields with constrained access anyway.
I've never found them to be terribly useful and I do think they blur the
lines a bit much. I find that it is too easy to consider a field as local
state, even with the static marker.
 
If I have table flight with flightno, from, to,.. I want my program to
return all possible connections ,given from X and to Y.
Here is the solution I come up with:
1.Search all flights from X
For example I get:

flightno from to
1 X Y
2 X A
3 X B

2.Loop thru the result
a. If the flight's destination is Y, return this path
b. Else, Search all flights goes from this result's location (for
example: all flights from A)

I hope I make it clearer? Now you see why I want to keep the old
datareader?
 
Kevin Spencer said:
Yrsh, it kind of sucks that you can't declare static method variables, but
there you are. Declare them at class scope, as private fields, instead.
Click to expand...

Personally, I'm glad that static method variables don't exist. If a
method needs that much state (so it's hard to pass as parameters
recursively, for instance), it's usually a sign that that state should
be encapsulated in a class.
 
Personally, I'm glad that static method variables don't exist. If a
method needs that much state (so it's hard to pass as parameters
recursively, for instance), it's usually a sign that that state should
be encapsulated in a class.
Click to expand...

That can be argued. Still, I like to have the freedom. Hey, I'm something of
a control freak!

Anyway, note that I only said "kind of." I'm not losing any sleep over it!

--

Kevin Spencer
Microsoft MVP
..Net Developer
If you push something hard enough,
it will fall over.
- Fudd's First Law of Opposition
 
Kevin Spencer said:
That can be argued. Still, I like to have the freedom. Hey, I'm something of
a control freak!
Click to expand...

In that case, C++ is the language for you. C# stops you from doing all
kinds of things which are usually a bad idea - and as I say, I'm glad
about it.

Philosophically, methods are actions - they don't have state when
they're not on the stack. Objects and types *do* have state, and that's
where C# forces you to put that state.
 
Philosophically, methods are actions - they don't have state when
they're not on the stack. Objects and types *do* have state, and that's
where C# forces you to put that state.
Click to expand...
I like static method var just because it frees me from worrying it
might be unintentionally accessed by something else. When I use a
method static var, what I actually want is an object state to which
access is restricted to a single method. If there is something like a
compile-time attribute that I can apply to a field to instruct the
compiler to emit a compile-time error if the field is accessed not by
one of the methods specified, no-one would use method static vars then.
That is my idea.

Thi
 
Truong said:
I like static method var just because it frees me from worrying it
might be unintentionally accessed by something else. When I use a
method static var, what I actually want is an object state to which
access is restricted to a single method. If there is something like a
compile-time attribute that I can apply to a field to instruct the
compiler to emit a compile-time error if the field is accessed not by
one of the methods specified, no-one would use method static vars then.
That is my idea.
Click to expand...

While I can see that being useful, I think it's worth considering *why*
you want it to only be available to that method. Is it *fundamentally*
part of the state of the object (or type for a static method) or is it
really state which perhaps belongs somewhere else?

I know it's not always clear-cut, but I tend to find that when there
are variables like this, a redesign gives a more elegant solution.

Jon
 
Hi Jon,
In that case, C++ is the language for you. C# stops you from doing all
kinds of things which are usually a bad idea - and as I say, I'm glad
about it.
Click to expand...

Nah. I prefer the ease of use with C#, the elegance of the syntax, and I can
live with the managed compromises. I was just wishing out loud. I can
certainly understand the reasons for the constraints. In this biz,
everything involves a certain amount of compromise. If I couldn't live with
that, I'd go crazy!

The number of possible permutations involved in the process of skinning a
cat are nearly infinite, and finding an elegant cat-skinning method is a
challenge I enjoy. :)


--

Kevin Spencer
Microsoft MVP
..Net Developer
If you push something hard enough,
it will fall over.
- Fudd's First Law of Opposition
 
Hi Jon,

One other point I couldn't get off my mind:
Philosophically, methods are actions - they don't have state when
they're not on the stack. Objects and types *do* have state, and that's
where C# forces you to put that state.
Click to expand...

This is a bit of an over-simplification. Methods have both state and
process. Variables in methods are certainly state, not process. So, the
difference betweeen methods and classes is more subtle that you seem to
propose. Still, as I mentioned earlier, compromises must always be made.
Lines must eventually be drawn somewhere. I am satisfied with the places
where Microsoft has drawn the ines in .Net. The important issue to me is,
can I do what I need to do with relative ease and high productivity using
these tools? And of course, the answer is a resounding "yes."

--
:-D,

Kevin Spencer
Microsoft MVP
..Net Developer
If you push something hard enough,
it will fall over.
- Fudd's First Law of Opposition
 
Kevin Spencer said:
One other point I couldn't get off my mind:


This is a bit of an over-simplification. Methods have both state and
process. Variables in methods are certainly state, not process.
Click to expand...

Yes - but only when they're on the stack (i.e. either running or ready
to run when a method they've called returns).
So, the difference betweeen methods and classes is more subtle that
you seem to propose.
Click to expand...

Not with the difference above - a method which is not executing (and
nowhere on the stack) doesn't have any state, which is consistent with
real life - an action itself doesn't tend to have state apart from when
it's happening. (If I'm "going to the shop" I know where I happen to
be, but the concept of "going to the shop" itself doesn't have state.)
Still, as I mentioned earlier, compromises must always be made.
Lines must eventually be drawn somewhere. I am satisfied with the places
where Microsoft has drawn the ines in .Net. The important issue to me is,
can I do what I need to do with relative ease and high productivity using
these tools? And of course, the answer is a resounding "yes."
Click to expand...

Gooodo :)
 
Not with the difference above - a method which is not executing (and
nowhere on the stack) doesn't have any state, which is consistent with
real life - an action itself doesn't tend to have state apart from when
it's happening. (If I'm "going to the shop" I know where I happen to
be, but the concept of "going to the shop" itself doesn't have state.)
Click to expand...

Neither does a class which is not instantiated (and nowhere on the stack).

--
;-),

Kevin Spencer
Microsoft MVP
..Net Developer
If you push something hard enough,
it will fall over.
- Fudd's First Law of Opposition
 
Kevin Spencer said:
Neither does a class which is not instantiated (and nowhere on the stack).
Click to expand...

The type itself can have state, although as we've said elsewhere, it's
generally not a good idea to have publicly mutable state. However, the
Process type effectively has "state" in terms of "the currently running
processes" - I don't see anything inconsistent in that.
 
Back
Top