[2221] in java-interest
java-interest-digest V1 #186
daemon@ATHENA.MIT.EDU (owner-java-interest-digest@java.su)
Wed Sep 27 05:27:17 1995
From: owner-java-interest-digest@java.sun.com
Date: Mon, 25 Sep 1995 17:07:27 -0700
To: java-interest-digest@java.sun.com
Reply-To: java-interest@java.sun.com
java-interest-digest Tuesday, 26 September 1995 Volume 01 : Number 186
In this issue:
regular expressions and dbm for Java?
Re: A couple of things...
Re: One File for all
Public class == File Name = Why?
unsuscribe
Re: Fn ptrs & ... callbacks!?
Consoles and repaints
Re: regular expressions and dbm for Java?
Re: Consoles and repaints
modulus on floats?
what's become of pushURL in JDK ?
native library path
JOE & Sun's position on HotJava ??
Re: JOE & Sun's position on HotJava ??
Re: Q's on File access API's in Pre-Beta
interfaces are NO substitution for method pointers
Re: Object level concurrency control.
Re: Wouldn't it be nice if...
class loaders
Re: interfaces are NO substitution for method pointers
Re: throws declataration in Java/beta
Re: native method build problem
Re: inadequate base classes
Re: class loaders
Re: class in differents files
Re: WWWClassLoader problems (mobile objects)
Re: class loaders
----------------------------------------------------------------------
From: tmb@best.com
Date: Sun, 24 Sep 1995 23:56:40 -0800
Subject: regular expressions and dbm for Java?
Languages like Perl and Python that are currently popular for doing
a lot of WWW related programming have extensive support for
regular expressions and dbm files.
I'm sort of missing similar support in Java. Is there a standard
API and implementation in the works?
Thomas.
------------------------------
From: "Elliotte Rusty Harold" <Elliotte@blackstar.com>
Date: Mon, 25 Sep 1995 08:53:29 EST
Subject: Re: A couple of things...
> 4. I love HotJava. I think it's well specified for the important
> niche of applications-for-the-Web. Yes, it's a shame the docs aren't
> any better. I guess we all learned Java by looking at other people's
> sources. It seems to be a pitfall of doc-generating programs to end
> up with lots of *documentation* holding very little *information*.
> Has Sun people working on the docs?
Yes. There are also several independent efforts including mine at
providing more java documentation.
- --
Elliotte Rusty Harold Black Star Publishing Co., Inc.
elliotte@blackstar.com 116 East 27th Street
elharo@escape.com NY NY 10016
------------------------------
From: ajg@cadre.com (Andrew Gilmartin)
Date: Mon, 25 Sep 95 10:00:53 EDT
Subject: Re: One File for all
>It would be good to pack all files that belong to an applet
>into one file. This would allow us to transfer the data packed
>via the net and unpack it at cilent side.
>
>Is it possible to programm such a fileType in Java???
Checkout Jarchive at "http://www.cs.brown.edu/people/amd/java/jar/".
- -- Andrew
- --
Andrew Gilmartin
Cadre Technologies Inc.
ajg@cadre.com
------------------------------
From: anders.rundgren@seuppbt.pharmacia.se
Date: Mon, 25 Sep 1995 15:14:52 +0100
Subject: Public class == File Name = Why?
Hi,
In the current (Alpha3) language specification there is a line that says=20
that there can be max 1 public class or interface per CU (compilation Unit)=
.=20
Also there is a compiler warning if the source file differs from the name=20
of the public class. The reason behind all this seems to be that the=20
compiler can automatically invoke source files for external classes.
My question is: Is Java the only computer language where the source file=20
name is a part of the language grammar? Should not be there IMO.
Assuming that I am correct this is what I would like to see in the Beta=20
release:
1. Delete the requirement that multiple public classes cannot be in the sam=
e=20
CU. It is nice to put public classes and interfaces that belong together i=
n=20
the same file. BTW, It works just fine in A3. You get a warning but thats=20
all. In the planned Beta all warnings have been taken away so this pratice=
=20
is soon broken.
2. As the auto-source-search feature *may* be useful (until an integrated=20
environment is here) if standard MAKE files are to be used, this should be=20
controllable with a switch like "-disable-auto-source-search" that=20
effectively disable the tests for multiple public classes/interfaces and=20
source file name comparisions.
Anyone, right or wrong?
Regards
Anders Rundgren
Vincent Engineering
------------------------------
From: david@moorea.llnl.gov (David Ashkenas)
Date: Mon, 25 Sep 95 07:17:16 PDT
Subject: unsuscribe
unsuscribe
------------------------------
From: Sal Cataudella <salc@ovid.com>
Date: Mon, 25 Sep 1995 11:45:03 -0400 (EDT)
Subject: Re: Fn ptrs & ... callbacks!?
On Sat, 23 Sep 1995, Gary Aitken wrote:
> >> Let's say you need to create 2 sliders, and have each one of them invoke
> >> a different method. With the method ptr callback technique, you can create
> >> 2 instances of sliders and simply register the method and object pointers
> >> you wish to have invoked. Thus , you are reusing code (the slider's)
> >> by instantion, not by subclassing. Without the method callback technique,
> >> you are *forced* to subclass *twice*, in order to reuse code.
> >
> >You could define a new interface, Callbackable that defines a
> >callback method. Then create a single slider subclass that
> >manages a Vector of Callbackable instances. In the sliders
> >callback method, call the callbackable method of each
> >Callbackable in the Vector.
> >
> >So now you can give every slider one or more different
> >Callbackable objects that implement your callback functionality.
> >Your callback functionality is independent of your slider class.
> >
> >If you change all your UI components that use callbacks (Button
> >etc.) to manage a Callbackable Vector, then you can even share
> >Callbackable instances among different UI instances (e.g. I write
> >a Callbackable that pops up some dialog, I can stick that in a
> >menu item and a button - when either one is clicked it will
> >execute the same shared code and pop up a dialog).
> >
> >I don't see how this is worse than function pointer callbacks, I
> >prefer it. In C you typically need to register some void* pointer
> >along with your function pointer to give your function some
> >context. With Callbackable, you're encapsulating that data in
> >the Callbackable object in a typesafe way.
>
>This is all true. However, it would be far better if the callblackable vector
>were part of the base classes. If you think of code reuse and general systems
>which want to be able to deal with arbitrary objects, the last thing you want
>to have to do is derive a new class for every darn class someone writes.
Actually, with the Callbackable vector scheme you will be subclassing
for *every* method which you want to have called back, not just every
new class which you want called back.
Again, polluting of the class name space would be avoided with a builtin
method reference type.
> Gary Aitken garya@village.org
> -
> Note to Sun employees: this is an EXTERNAL mailing list!
> Info: send 'help' to java-interest-request@java.sun.com
>
Sal Cataudella email: salc@ovid.com
Ovid Technologies (formerly CD Plus) phone: 212-563-3006 ext. 406
333 Seventh Ave., 4th Floor fax: 212-563-3784
New York, NY 10001
------------------------------
From: "Elliotte Rusty Harold" <Elliotte@blackstar.com>
Date: Mon, 25 Sep 1995 12:07:01 EST
Subject: Consoles and repaints
Two questions that aren't apparently related:
1. In HotJava alpha 2 for NTand apparently on Solaris
there was always a console window open to receive the results of
commands like System.out.println and error messages. That apears to
have gone away in HotJava alpha 3 for NT. Is there a way to get it
back?
2. Is there a method I can call to force an applet to repaint?
- --
Elliotte Rusty Harold Black Star Publishing Co., Inc.
elliotte@blackstar.com 116 East 27th Street
elharo@escape.com NY NY 10016
------------------------------
From: Michael Lorton <mlorton@eshop.com>
Date: Mon, 25 Sep 1995 10:26:55 -0700
Subject: Re: regular expressions and dbm for Java?
> Languages like Perl and Python that are currently popular for doing
> a lot of WWW related programming have extensive support for
> regular expressions and dbm files.
>
> I'm sort of missing similar support in Java. Is there a standard
> API and implementation in the works?
Reminds me so say some things I have been meaning to say for a while.
1. If there is no available API, and you are planning to do an
implementation, write it as an *interface*. That way, people can use
your implementation without becoming tethered to it.
2. Sun ought to have an informat interface-approval process (based, I
suppose, on RFCs). Standardize database, R-E, encryption, what have
you.
M.
------------------------------
From: jpayne@starwave.com (Jonathan Payne)
Date: Mon, 25 Sep 1995 10:37:15 -0700
Subject: Re: Consoles and repaints
> Two questions that aren't apparently related:
>
> 1. In HotJava alpha 2 for NTand apparently on Solaris
> there was always a console window open to receive the results of
> commands like System.out.println and error messages. That apears to
> have gone away in HotJava alpha 3 for NT. Is there a way to get it
> back?
Just run hotjava from a console window to begin with:
java browser.hotjava
> 2. Is there a method I can call to force an applet to repaint?
repaint()
------------------------------
From: lemay@lne.com (Laura Lemay, Killer of Trees)
Date: Mon, 25 Sep 1995 10:57:20 -0800
Subject: modulus on floats?
I've been playing with arithmetic (I'm one of the unwashed hordes
writing books).
The JDK Change Notes say "the meaning of floating point % got clarified."
To what?
The alpha spec says that a % b on floats is equivalent to:
a - ((int)(a / b) * b)
However, that's not what a % b is currently behaving like, and I'm getting
really bizarre results. 12.5 % 7.5, for example, is now -1.5. Huh?
What's the new formula for floating-point modulus?
Laura
------------------------------
From: dufourd@enst.enst.fr
Date: Mon, 25 Sep 95 19:15:49 +0100
Subject: what's become of pushURL in JDK ?
Not long ago, I have seen some exchange about the JDK equivalent
of pushURL, but I did not keep the mail, and the mail archives do not
seem to have anything newer than from last month ... so ...
Where has pushURL gone ?
It seems, from reading the docs, that there should be an
URL.getContent() somewhere, but will that be enough ?
JCD
- ------------------------------------------Java, Lisp & Theater----------------
Jean-Claude Dufourd Departement Electronique,ENST Tel: +33 1 45 81 78 07
46 rue Barrault, 75013 Paris Fax: +33 1 45 80 40 36
Email: dufourd@elec.enst.fr URL: http://www-elec.enst.fr/~dufourd/jcd.html
------------------------------
From: downing@GRAPHICS.CS.NYU.EDU (Troy Downing)
Date: Mon, 25 Sep 95 14:55:32 EDT
Subject: native library path
I've been working on developing Native libraries for
a while under Solaris. When I use the System.loadLibrary("foo");
call it looks for libfoo.so in the path set in LD_LIBRARY_PATH.
Question: what is the Windows NT equivalent to the
LD_LIBRARY_PATH environment variable in UNIX.
- -Troy
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Troy Downing, Research Scientist (Voice) (212) 998-3208
New York University (FAX) (212) 995-4122
Media Research Lab
715-719 Broadway, Rm 1214 downing@nyu.edu
New York, NY 10003-1866 http://found.cs.nyu.edu/downing
------------------------------
From: eric@internet.sbi.com (Eric Ho)
Date: Mon, 25 Sep 95 18:28:16 EDT
Subject: JOE & Sun's position on HotJava ??
Hi there,
From last week's Java Day in NYC, I heard some rumors about JOE (DOE in Java).
Is this for real ??
Also, I'm curious as to what's Sun's position in HotJava. Is Sun planning to
pour money into this browser when over 70% of the browser market is owned by
Netscape (plus the fact that Navigator 2.0 will have Java inside) ?
- --
Best regards.
- -Eric Ho- eric@sbi.com, http://rutherford2.sbi.com:2020/eric/
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 16:03:34 -0800 (PDT)
Subject: Re: JOE & Sun's position on HotJava ??
Hi Eric,
> From last week's Java Day in NYC, I heard some rumors about JOE (DOE in
> Java). Is this for real ??
Yes it is real. Contact Greg Nuyens (greg.nuyens@eng.sun.com) for more info.
Have fun,
Arthur van Hoff
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 16:14:28 -0800 (PDT)
Subject: Re: Q's on File access API's in Pre-Beta
Hi Bob,
> In browsing the pre-beta API pages, it looks like some of the issues I've
> had with alpha3 haven't been addressed:
>
> 1) No way to delete a file
> 2) RandomAccessFile doesn't have a getFD() method (FileInputStream
> does, though)
Thanks for pointing this out. I'll try to make sure they get added. I'm not
sure how you
> 3) There doesn't appear to be any API for locking files or portions
> of files
I'm not sure how we implement this in a platform independent way.
> If these are there, please let me know (I'm running on NT, thus can't use
> the pre-beta). If these aren't there, when will they (or some such support)
> - it'll be hard to argue general purpose programs can be written in Java
> without these.
Have fun,
Arthur van Hoff
------------------------------
From: jpayne@starwave.com (Jonathan Payne)
Date: Mon, 25 Sep 1995 12:33:02 -0700
Subject: interfaces are NO substitution for method pointers
OR, somebody please tell me why I am confused.
It has been suggested that using interfaces is a substitution for
method pointers. How's that? How would I implement EMACS keymaps
using interfaces? Or, if not interfaces, a class with an action
method, that you subclass for each callback.
I would have to create N classes for each entry in an Emacs keymap
that runs a difference command. I really don't want to create N
classes for N callbacks, especially when EACH OF those callbacks
REALLY wants to be executing in the context of another class (say, the
EditorBuffer class). No thanks. I want method pointers.
Is there a security issue that I am not understanding?
Regardless of whether Java is going to have method pointers some time
in the future, please don't tell me that interfaces or "action"
classes are a good substitution. That's ludicrous.
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 14:31:46 -0800 (PDT)
Subject: Re: Object level concurrency control.
Hi Leon,
> Seen as Java is both multithreaded and Object Oriented I was wondering
> if there was any object level concurrency control available or in the
> works. For example can one specify a set of methods as "serialised",
> meaning that only one of these methods can be executing at a time.
That is essentially what "synchronized" does.
> The reason I ask is wrt to object migration. In a multithreaded
> environment it might not be a good idea to migrate an object while one
> of its methods is executing! Hence one might might be able to make the
> its migrate method a member of this group and so avoid the possibility
> of migrating as one of its (public) methods is being executed by a
> different thread.
We don't support any form of object migration or persitification other
than what you can hack yourself.
Have fun,
Arthur van Hoff
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 14:07:45 -0800 (PDT)
Subject: Re: Wouldn't it be nice if...
Hi Russell,
> I just learned that putting an Applet thread to sleep in the wrong
> place can block other Applets of the same kind. For example:
>
> // ...
> Thread life;
> // ...
> public void paint() {
> for (int i = 0; i < 100; i++) {
> // ...
> life.sleep(50);
> }
> }
>
> So, I guess we're getting one event queue per Browser and not one
> per Applet? Wouldn't it be nice if each Applet had it's own event
> queue? Is this just a Windows NT/95 thing??
>
> Or, is paint() synchronized and one Applet instance doing a paint()
> locks out other Applet instances from entering paint()??? Hope not.
>
> When the real doc's come out, I hope we'll be able to read in great
> depth about how user events turn into method calls; which methods
> are event "handlers" and which aren't; which are synchronized, etc.
There is one special thread called the "AWT Thread" which is needed
to handle window system events. If you block that thread, it will
affect all other applets.
Have fun,
Arthur van Hoff
------------------------------
From: jpayne@starwave.com (Jonathan Payne)
Date: Mon, 25 Sep 1995 15:43:56 -0700
Subject: class loaders
Can I define my own class loader which loads classes from the file
system? If I do, will I be subject to the security measures which are
in place for WWWClassLoader, which loads classes from the network? Or
are the security measures dependent on that particular class loader
subclass?
Also, if my class loader periodically checks the file system for an
update to that class, and reloads that class, will the system actually
successfully start using the new definition of the class. Something
tells me that it will actually continue to use the original
definition, because the class is actually stored internally in the
interpreter.
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 16:31:44 -0800 (PDT)
Subject: Re: interfaces are NO substitution for method pointers
> OR, somebody please tell me why I am confused.
>
> It has been suggested that using interfaces is a substitution for
> method pointers. How's that? How would I implement EMACS keymaps
> using interfaces? Or, if not interfaces, a class with an action
> method, that you subclass for each callback.
>
> I would have to create N classes for each entry in an Emacs keymap
> that runs a difference command. I really don't want to create N
> classes for N callbacks, especially when EACH OF those callbacks
> REALLY wants to be executing in the context of another class (say, the
> EditorBuffer class). No thanks. I want method pointers.
>
> Is there a security issue that I am not understanding?
>
> Regardless of whether Java is going to have method pointers some time
> in the future, please don't tell me that interfaces or "action"
> classes are a good substitution. That's ludicrous.
I don't think it is ludicrous, interfaces are strongly typed sets of methods.
Anyway, there seems to be no point in trying to convince you so I won't.
Have fun,
Arthur van Hoff
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 16:33:32 -0800 (PDT)
Subject: Re: throws declataration in Java/beta
Hi,
> >This means that you can NEVER throw an exception from a
> >toString() method because Object.toString() doesn't throw
> >any exceptions.
>
> This is my experience too, and this language/compiler design decision
> seems pretty lame to me. Another thing you cannot do is ask for the
> current thread to sleep when you are in the run method because the
> interface Runnable does not specify "throws InterruptedException"...
> I am sure there are lots of other such examples.
It is very important to realize that a Java class is a guaranteed interface.
If an method is declared to throw a certain set of exceptions than there is no
way a sub class can change that. At first this may seem restrictive, but after
a while you will realize that it is wonderful to have this guarantee.
Have fun,
Arthur van Hoff
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 16:15:20 -0800 (PDT)
Subject: Re: native method build problem
Hi Jon,
> I have been trying to implement a native method and was having trouble.
> I went back to the demo in the alpha3 release ..../hotjava/doc/native
> where there is a Makefile and typed "make"
> I got the following...
> ------------------------------------------------------------------------
> javac demo/File.java
> javac demo/InputFile.java
> demo/InputFile.java:42: Class or interface definition expected.
> package demo;
> ^
> demo/InputFile.java:52: Super class File of class InputFile not found.
> class InputFile extends File {
> ^
> 2 errors
> *** Error code 1
> make: Fatal error: Command failed for target `InputFile.class'
> ------------------------------------------------------------------------
>
> Am I doing something silly? Is there a simple fix?
You need to import java.io.File.
Have fun,
Arthur van Hoff
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 16:39:02 -0800 (PDT)
Subject: Re: inadequate base classes
Hi George,
> My company has been looking at using Java for
> a substantial project. The only reason I have found
> not to use Java is that the awt class does not
> meet my minimum needs. Those needs include
> assess to a second mouse button and to some
> function keys.
>
> I am sure that other potential users will have
> the same problems. My options appear to be:-
>
> 1. Have the Java team extend the awt class.
> Would be my best answer but could result in Java
> not being runable on all platforms. I am sure
> that I will not get this result.
>
> 2. Have the Java team provide the extensions I
> need in a way that they could be disabled by
> 'pure Java' applets, but enabled in 'contaminated
> Java' programs.
>
> 3. Put an extended awt class into the sun class
> tree. At least this would provide a standard
> for those who need the additional capabilities.
>
> 4. Build my own copy of the awt class. And so will
> many other programmers. I will get what I need,
> but at the cost of damaging the ability of
> developers to share code. WE NEED TO AVOID THIS
> IF WE CAN.
>
> What I really need is some comment from the
> Java team about these problems.
The right thing to do seems a combination of 1 and 3.
We need some additional functionality to implement the
HotJava browser and we are putting that in sun.awt.
I think that we'll at least add some function key events.
That seems reasonably easy to support on all platforms.
We can also add support for 1, 2 and 3 button mice as long
as there is some well defined way to invoke these from
a lesser mouse.
Have fun,
Arthur van Hoff
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 16:41:49 -0800 (PDT)
Subject: Re: class loaders
Hi Jon,
> Can I define my own class loader which loads classes from the file
> system?
In a normal Java program yes, but not from an applet.
> If I do, will I be subject to the security measures which are
> in place for WWWClassLoader, which loads classes from the network?
Yes.
> Or are the security measures dependent on that particular class loader
> subclass?
You can't do it from an applet so this is not really an issue.
> Also, if my class loader periodically checks the file system for an
> update to that class, and reloads that class, will the system actually
> successfully start using the new definition of the class. Something
> tells me that it will actually continue to use the original
> definition, because the class is actually stored internally in the
> interpreter.
The next time that the class is loaded from that class loader it will
return the new class. The old class will eventually be gc'd (although
that isn't implemented). Existing instances of the new class will not
automatically be upgraded.
Have fun,
Arthur van Hoff
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 16:43:03 -0800 (PDT)
Subject: Re: class in differents files
Hi Bruno,
> I would like writing a package with several files (.java). And some of them
> contains some objects of a class which is writting in an other file and this
> file contains some objects which have their class type in the first file.
>
> For exemple :
>
> ------------------
> //classA.java
>
> package MyPackage;
>
> class ClassA
> {
> ClassB cb;
> ...
> }
>
> ------------------
> //classB.java
>
> package MyPackage;
>
> class ClassB
> {
> ClassA ca;
> ...
> }
>
> In this way if I compile classA.java, it don't know classB and vis and versa.
> If someone have any idea to do this ? (classA.java and classB.java are
> compiling in the same directory).
>
> -------------------------------------------------
You need to put your classes in a subdirectory called "MyPackage". For example:
mkdir MyPackage
mv classB.java classA.java MyPackage
javac MyPackage/*.java
Have fun,
Arthur van Hoff
------------------------------
From: Arthur van Hoff <Arthur.Vanhoff@Eng.Sun.COM>
Date: Mon, 25 Sep 1995 16:45:33 -0800 (PDT)
Subject: Re: WWWClassLoader problems (mobile objects)
Hi Chanda,
> I'm trying sample code to load a class from InputStream.
> Following a couple of postings on this mailing list I tried using the
> WWWClassLoader definitions.
>
> - If I understand correctly loadClass(InputStream is) is defined in
> WWWClassLoader.java.
In alpha3 you are right.
> - This reads the input buffer and calls defineClass(...)
I think so.
> - defineClass does resolution and further "native" stuff which I assume
> is generating the tree and checking for violations etc..
Right again.
> 1) Have I understood this correctly ? So if the referenced classes are
> also on remote machines I have to make sure they are all loaded
> before the current one, by explicitly calling loadClass ??
The system automatically resolves referenced classes by making calls
to loadClass(String).
> 2) I'm having other problems with WWWClassLoader. However, before
> posting any questions I wonder if anyone has saved a copy of Chuck
> McManis's "Secrets of the ClassLoader" ? Or any other pointer to
> information that may be useful in understanding this buisness..
Have fun,
Arthur van Hoff
------------------------------
From: cmcmanis@scndprsn.Eng.Sun.COM (Chuck McManis)
Date: Mon, 25 Sep 1995 17:05:12 -0700
Subject: Re: class loaders
>Can I define my own class loader which loads classes from the file
>system?
Yes.
> ... If I do, will I be subject to the security measures which are
>in place for WWWClassLoader, which loads classes from the network?
In alpha3 yes.
> ... Or
>are the security measures dependent on that particular class loader
>subclass?
In Beta the security measures are implemented by a "SecurityManager".
There is a subclass of SecurityManager called AppletSecurity that defines
the policies for applets.
If you are an applet, the AppletClassloader will not let you instantiate
a new class loader. If you are simply a Java program you have the option
to call the security manager in your loadClass method if you so choose.
>Also, if my class loader periodically checks the file system for an
>update to that class, and reloads that class, will the system actually
>successfully start using the new definition of the class. Something
>tells me that it will actually continue to use the original
>definition, because the class is actually stored internally in the
>interpreter.
Your latter statement is the more correct one. It is of course possible
to define a native method in your class loader that does a manual substitution
on the internal class list however that is probably a fairly dangerous
operation, especially if you make an incompatible class change.
- --Chuck
------------------------------
End of java-interest-digest V1 #186
***********************************
-
Note to Sun employees: this is an EXTERNAL mailing list!
Info: send 'help' to java-interest-request@java.sun.com
