aglets-developer — List for Aglets platform developers.

RE: [Aglets-developer] Easy Installation of ASDK?

[Thomas C. <cal...@te...>]

Todd

Regarding setting up AGLET_PATH, etc.:

"it would probably save a lot
of confusion in the long run."

Agreed.

" My reading of the GPL just means that if commercial orgs want
to use the installer they have to release any changes they made to
that installer, but since it's seperate from the main ASDK that's all it
will affect."

The separation might solve the licensing issue, but it becomes a question of
whether the installer is indeed separate from the application it is taken
from and whether it is considered a library of the application it is used
with. If considered a library, then the GPL gets, well, complicated. I am
actively researching what it will take to do a quick and easy installation.
If it looks like a big hassle, I will check into the jEdit code and we can
try to figure out the licensing issues from there.

Tom

Re: [Aglets-developer] Re: Who sent the message?

[Marc M. <ma...@co...>]

I'm just throwing a few ideas out right now....   BTW, yes, I
think your use case captures the basic process.

1) Should aglets or contexts or both be able to say they will only
accept authenticated messages?  Can an aglet tell the context to
never send out information about it, if it wants to stay anonymous
(or at least as anonymous as possible)?

2) If an entity requires authenticated messages and it receives an
ordinary message, should there be a challenge/response mechanism?
Should the server be able to send a reply to the message asking
for authentication, kind of like HTTP basic authentication? 

3) It seems rather straight forward (if not easily implemented) to
enforce trust at the inter-server level.  The Tahiti operator imports
the certificates he wants (e.g. Verisign).  What happens if an
aglet wants to trust messages from a source that the context does
not trust?

4) I think a public key cryptosystem -- as opposed to shared keys --
makes more sense.  It is pretty easy to code in Java once you
get the feel for java.security.cert.  The only problem is getting
a common base of trusted certificate authorities.  Since the Java
keystore does not ship with anything, there would need to be a
bootstrap process.

5) One form of trust that I am investigating is tying an Aglet's
package name to the codesigner's certificate name.  For instance,
if I want to sign an Aglet named "edu.ucsc.cse.Mosko", I would need
a certificate with a subject like "CN=ucsc.edu, ...".

6) How would this authentication relate to something like ATP/HTTP
over SSL?  SSL has the concept of server and client authentication
via certificates.  Would Tahiti servers be able to re-use an SSL
certificate for message signing?  If you go with a Verisign cert,
for instance, you're paying $$ every year, so you'd want to buy
as few as you could.

7) If an aglet says that it will accept non-authenticated messages
and the context receives a message that fails authentication, what
should happen?

8) One issue to be addressed is that if the context will sign messages
attesting to an aglet's identity, the context needs to be sure of 
that identity.  Otherwise, you get something similar to the
"cross site scripting" security problem of some web browsers.  What
configuration parameters are needed here?  This will get in to a
tricky system of belief and the interactions of systems with different
concepts of trust.

Hope that didn't muddy the waters too much,
Marc



> Robert Bergstrom wrote:
> 
> Right, the only private key currently available would be that of the
> context owner, so it could sign any information necessary.  A
> prerequisite, would be establishing a trust
> 
> domain between the two servers which could be done with the shared
> secret mechanism.
> 
> So, the first use case is that an aglet running in a trusted context
> sends a message to and aglet running in another (same?) context.
> Then, assuming the sending aglet has the proper message permissions in
> the context, the context would compile the info on the aglet and place
> it and a signature in the message object.  Upon reciept, the recieving
> context verifys the signature, and if satified dispatches the message
> to the target aglet.  You'd want to use same mechanism for any
> replies.  Should there be any notification of rejection to the sender
> aglet?  Did I miss anything?
> 
> At what granularity did you want to grant permissions?  Codebase
> signature?  I like the idea of the context doing the checks.  That way
> the power to dispose of an aglet lies with the server owner.
> 
> Thoughts?
> 
> // RAB
> 
> P.S. How would the servers respond to HTTP messaging or is that not a
> concern?
> 
> > -----Original Message-----
> > From: Marc Mosko [mailto:ma...@co...]
> > Sent: Thursday, March 15, 2001 12:43 PM
> > To: aglets-developer
> > Subject: [Aglets-developer] Re: Who sent the message?
> >
> >
> > I have moved this thread from Aglets-user to Aglets-developer at
> > Robert's suggestion.
> >
> > It seems to me that the only way to get some modicum of security is
> > to have Tahiti sign the messages and attest to the class name of the
> 
> > executing agent.  It could also include the codebase signature.  I
> > do not think it would work for the agent itself to try to say it's
> > class name.  It could just lie.  Or, someone outside the whole
> Tahiti
> > environment could forge a message.  Also, since it's not a good
> > idea to put private keys in agents, the agents can't really sign
> > anything themselves.
> >
> > Marc

RE: [Aglets-developer] Re: Who sent the message?

[Robert B. <rbergstrom@iSolutionsCorp.com>]

Right, the only private key currently available would be that of the context
owner, so it could sign any information necessary.  A prerequisite, would be
establishing a trust
domain between the two servers which could be done with the shared secret
mechanism.

So, the first use case is that an aglet running in a trusted context sends a
message to and aglet running in another (same?) context.  Then, assuming the
sending aglet has the proper message permissions in the context, the context
would compile the info on the aglet and place it and a signature in the
message object.  Upon reciept, the recieving context verifys the signature,
and if satified dispatches the message to the target aglet.  You'd want to
use same mechanism for any replies.  Should there be any notification of
rejection to the sender aglet?  Did I miss anything?

At what granularity did you want to grant permissions?  Codebase signature?
I like the idea of the context doing the checks.  That way the power to
dispose of an aglet lies with the server owner.

Thoughts?

// RAB

P.S. How would the servers respond to HTTP messaging or is that not a
concern?


> -----Original Message-----
> From: Marc Mosko [mailto:ma...@co...]
> Sent: Thursday, March 15, 2001 12:43 PM
> To: aglets-developer
> Subject: [Aglets-developer] Re: Who sent the message?
> 
> 
> I have moved this thread from Aglets-user to Aglets-developer at
> Robert's suggestion.
> 
> It seems to me that the only way to get some modicum of security is
> to have Tahiti sign the messages and attest to the class name of the
> executing agent.  It could also include the codebase signature.  I
> do not think it would work for the agent itself to try to say it's
> class name.  It could just lie.  Or, someone outside the whole Tahiti
> environment could forge a message.  Also, since it's not a good 
> idea to put private keys in agents, the agents can't really sign 
> anything themselves.
> 
> Marc

[Aglets-developer] Re: Who sent the message?

[Marc M. <ma...@co...>]

I have moved this thread from Aglets-user to Aglets-developer at
Robert's suggestion.

It seems to me that the only way to get some modicum of security is
to have Tahiti sign the messages and attest to the class name of the
executing agent.  It could also include the codebase signature.  I
do not think it would work for the agent itself to try to say it's
class name.  It could just lie.  Or, someone outside the whole Tahiti
environment could forge a message.  Also, since it's not a good 
idea to put private keys in agents, the agents can't really sign 
anything themselves.

Marc

> Robert Bergstrom wrote:
> 
> Marc:
> 
> Looking through the message classes, I don't see any accessible way to
> get the sender's identity. Seems like a good idea, however.  If you
> have any thoughts on implementation, I'd encourage you to post them to
> the developer list. ;)  We could try and work it into a next release.
> 
> // RAB
> 
> P.S. I posted your bugs to the bug tracker.  I hope to get those in
> the next release as well.
> 
> > -----Original Message-----
> > From: agl...@li...
> > [mailto:agl...@li...]On Behalf Of
> > Marc Mosko
> > Sent: Wednesday, March 14, 2001 8:19 PM
> > To: 'Aglets List' (E-mail)
> > Subject: [Aglets-users] Who sent the message?
> >
> >
> > Is there a way to find out the proxy and/or class name of the sender
> 
> > of a message?
> >
> > What I really want is to trace a message to the source and see who
> > signed the codebase.  I will then authorize the aglet to request
> > certain things in the message.  For my specific case, it's okay
> > to assume that the sender & recipient are both in the same context.
> >
> > Thanks,
> > Marc
> >
> > e.g.:
> >
> > public class test extends Aglet {
> > public boolean handleMessage(Message msg) {
> >  // Who sent msg????
> > }
> > }
> >
> > _______________________________________________
> > Aglets-users mailing list
> > Agl...@li...
> > http://lists.sourceforge.net/lists/listinfo/aglets-users
> >

Re: [Aglets-developer] Transporting GUI's & AWT

[Todd P. <to...@da...>]

--On Friday, March 09, 2001 11:29 AM +0000 DT...@pi... wrote:

> I have successfully created a GUI to be mobile leaving only one problem.
> When the GUI gets to the hosting aglet server, the buttons within the GUI
> no longer work.

You might want to take a look at the Portal Packages, which have an
example of how to take a window with you. You can get them from
the Aglet Portal at http://www.aglets.org

Todd

-- 
 Todd Papaioannou @ Earth  http://www.luckyspin.org
 Luckyspin @ TerraFirmA   http://terrafirma.terra.mud.org
 Why search for life when you can create it? http://www.dalilab.com
.         .        .       .      .     .    .   .  .  . .

RE: [Aglets-developer] Threads in an aglet so slow!

[Robert B. <rbergstrom@iSolutionsCorp.com>]

I've noticed this as well, but haven't had the bandwidth to profile the
system.  If you're under windows and just have a machine name in your URL
(i.e. atp://mymachine), you could be seeing the DNS timeout.  Just a
hypothesis.

// RAB

> -----Original Message-----
> From: DT...@pi... [mailto:DT...@pi...]
> Sent: Friday, March 09, 2001 7:14 AM
> To: agl...@li...
> Subject: [Aglets-developer] Threads in an aglet so slow!
> 
> 
> Can anyone tell me why are threads running under the aglet 
> server so slow?
> Compare this with the same threads running under a debugger 
> and there is an
> enormous difference. Is there any way to adjust this? My 
> aglet owns threads
> that communicate with a back end prog through RMI. It takes 
> so long for RMI
> negotiation to happen, much slower than under the debugger.
> 
> Any suggestions?
> 
> Thanks,
> Dave
> 
> 
> _______________________________________________
> Aglets-developer mailing list
> Agl...@li...
> http://lists.sourceforge.net/lists/listinfo/aglets-developer
> 

RE: [Aglets-developer] Transporting GUI's & AWT

[Robert B. <rbergstrom@iSolutionsCorp.com>]

Dave:

My guess would be a serialization problem.  My approach has been to have the
aglet only maintain the "model" of the GUI and recreate it after dispatch.
Remember that only non-transient, serializable state can be sent.  The
aglet's stack is not sent.

// RAB

> -----Original Message-----
> From: DT...@pi... [mailto:DT...@pi...]
> Sent: Friday, March 09, 2001 6:29 AM
> To: agl...@li...
> Subject: [Aglets-developer] Transporting GUI's & AWT
> 
> 
> Hi all,
> 
> I have successfully created a GUI to be mobile leaving only 
> one problem.
> When the GUI gets to the hosting aglet server, the buttons 
> within the GUI
> no longer work.
> 

[Aglets-developer] Threads in an aglet so slow!

[<DT...@pi...>]

Can anyone tell me why are threads running under the aglet server so slow?
Compare this with the same threads running under a debugger and there is an
enormous difference. Is there any way to adjust this? My aglet owns threads
that communicate with a back end prog through RMI. It takes so long for RMI
negotiation to happen, much slower than under the debugger.

Any suggestions?

Thanks,
Dave

[Aglets-developer] Transporting GUI's & AWT

[<DT...@pi...>]

Hi all,

I have successfully created a GUI to be mobile leaving only one problem.
When the GUI gets to the hosting aglet server, the buttons within the GUI
no longer work.

Before transport, a listener was added to the button to take an action when
the button was pressed. This works fine before transport. A
System.out.println("Action performed") in the action listener shows that
this is working.

After transport however, you can press the button as many times as you like
but that action listener is not being called. No system.out messages, no
actions taken, zip!

Is this an issue with serialization of AWT classes? Has anyone else had a
problem with shipping a GUI class like this?

Many thanks for any of your replies,
Dave

[Aglets-developer] [ANN] ASDK 1.2.0 Developer Released

[Robert B. <rbergstrom@iSolutionsCorp.com>]

Agleteers:

I've built and packaged the code found in the aglets1_2 module using JDK
1.3. This release contains unmodified code which means it does have the user
defined message security bug.  I wanted everyone to have a common baseline
for development and to set the stage for future releases.

Please send bugs to the bug tracker.  Extra style points awarded for having
an aglet test case. :) As always, changes and patches are welcome and
encouraged.

// RAB a.k.a Kbd4Hire 

RE: [Aglets-developer] Artificial Intelligence

[Daryl B. <Da...@sc...>]

Dear Robert,
 
        Wow, my messages now have to be "edited for sanity". I feel so...
insane. :)
        Yes, using aglets.net for distributing genetic algoritms across
servers would be perfect. In addition, the distributing of the problem
across servers is not the only gain that I can forsee from using aglets. The
other would be distributing the actual gene pool across servers; so that
chromosomes would be somewhat "localized" to a set of genes. Also, if the
fitness function was similar (very similar) across servers -- and was local
to the servers and not the "genetic aglets" themselves -- then that would
help too. Why do all this? Because a large problem could be dispersed across
many smaller aglet servers. For example, a research site that had its own
set of test data (and thus its own fitness function) and its own gene pool
could breed a certain population of aglets, whereas another research center
could breed another population. By constantly allowing cross-breeding
between the populations, we would essentially have one big population, which
would be able to solve a problem better, in theory, because of the varied
data that it works from. The genes would pick up the effects of having been
through both research centers -- the solution would take into account both
sets of data.
        I don't know how well this would work. My worry is that the two
different fitness evaluations would cause problems. Unless of course there
was a myriad of fitness functions, which is its own can of worms. Having
fitness functions compete for "food" (genetic aglets to kill off due to
weakness) would be an interesting thing to try -- but this requires further
study at the conceptual level.
 
Sincerely,
 
        Daryl.
 
 

-----Original Message-----
From: Robert Bergstrom [mailto:rbergstrom@iSolutionsCorp.com]
Sent: Tuesday, March 06, 2001 4:29 PM
To: agl...@li...
Subject: RE: [Aglets-developer] Artificial Intelligence



Daryl: 

I'm a fan of genetic algorithms.  I think the infrastructure for creating
aglet "populations" would be a benefit to both the researcher and user
communities.  A genetic algorithm framework based on aglets would be an
great project!  The initial framework could use a fixed list of servers
until public aglet servers are ubiquitous.

Tom (Calivera) and I have made some very small steps toward creating the
infrastructure for a public network i.e. the so-called aglets.net.  We've
done some prototyping, but the 1.2 ASDK is not quite stable enough for
people to throw up public aglet servers.  That's where I'm focusing my
efforts, but I'd be glad to chime in on any "Genetic Aglet" discussion. :)

// RAB 

RE: [Aglets-developer] Artificial Intelligence

[Robert B. <rbergstrom@iSolutionsCorp.com>]

Daryl:

I'm a fan of genetic algorithms.  I think the infrastructure for creating
aglet "populations" would be a benefit to both the researcher and user
communities.  A genetic algorithm framework based on aglets would be an
great project!  The initial framework could use a fixed list of servers
until public aglet servers are ubiquitous.

Tom (Calivera) and I have made some very small steps toward creating the
infrastructure for a public network i.e. the so-called aglets.net.  We've
done some prototyping, but the 1.2 ASDK is not quite stable enough for
people to throw up public aglet servers.  That's where I'm focusing my
efforts, but I'd be glad to chime in on any "Genetic Aglet" discussion. :)

// RAB

-----Original Message-----
[Edited for sanity]
        When I came across Aglets, it lit a light in my mind. If a problem
was phrased in the open-source genetic algorithm language, it would be
possible to ask the "internet" (an aglets network) a question, and have it
eventually return an answer. It would also be possible for the agletnet to
evolve general solution algorithms, and thus contribute to human
understanding. What do you think? (that question goes to everybody on the
list!)

RE: [Aglets-developer] Easy Installation of ASDK?

[Robert B. <rbergstrom@iSolutionsCorp.com>]

Actually, Ant is already used for builds.  One goal for the installer is
that you could start with nothing except the jdk for running the jar file.
Ant's build files could be extended to create an installation, but you'd
first need to understand how to install Ant. :)

// RAB

> -----Original Message-----
> From: Daryl Beattie [mailto:Da...@sc...]
> Sent: Tuesday, March 06, 2001 2:07 PM
> To: 'agl...@li...'
> Subject: RE: [Aglets-developer] Easy Installation of ASDK?
> 
> 
> What about Apache's Ant? It is open-source too, I believe. [I 
> know nothing
> about it, beyond that it is a cross-platform make utility.]
> 
> http://jakarta.apache.org/ant/index.html
 
 

RE: [Aglets-developer] Easy Installation of ASDK?

[Daryl B. <Da...@sc...>]

What about Apache's Ant? It is open-source too, I believe. [I know nothing
about it, beyond that it is a cross-platform make utility.]

http://jakarta.apache.org/ant/index.html

> -----Original Message-----
> From: Todd Papaioannou [mailto:to...@da...]
> Sent: Tuesday, March 06, 2001 1:54 PM
> To: agl...@li...
> Subject: Re: [Aglets-developer] Easy Installation of ASDK?
> 
> 
> 
> 
> --On Tuesday, March 06, 2001 12:20 PM -0500 Thomas Calivera 
> <cal...@te...> wrote:
> 
> > What do you think? Is anyone already working on this? Does 
> anyone have code
> > snippets for installing from a executable JAR, creating a 
> certificate and
> > adding it to the keystore, etc.?
> 
> I think this is a great idea! Maybe the same installer could 
> ask people about
> the values of their AGLET_PATH etc at the same time. This has 
> always been
> one of the problems with aglets.
> 
> I don't have any code off hand, but one of the other OSS projects I'm
> associated with, jEdit (a great editor), has an java 
> installer that does
> a lot of neat things. The installer is also OSS, and could 
> probably be repurposed.
> I can find out where it can be downloaded from if you want.
> 
> Todd
> 
> -- 
>  Todd Papaioannou @ Earth  http://www.luckyspin.org
>  Luckyspin @ TerraFirmA   http://terrafirma.terra.mud.org
>  Why search for life when you can create it? http://www.dalilab.com
> .         .        .       .      .     .    .   .  .  . .
> 
> _______________________________________________
> Aglets-developer mailing list
> Agl...@li...
> http://lists.sourceforge.net/lists/listinfo/aglets-developer
> 

Re: [Aglets-developer] Easy Installation of ASDK?

[Todd P. <to...@da...>]

--On Tuesday, March 06, 2001 12:20 PM -0500 Thomas Calivera <cal...@te...> wrote:

> What do you think? Is anyone already working on this? Does anyone have code
> snippets for installing from a executable JAR, creating a certificate and
> adding it to the keystore, etc.?

I think this is a great idea! Maybe the same installer could ask people about
the values of their AGLET_PATH etc at the same time. This has always been
one of the problems with aglets.

I don't have any code off hand, but one of the other OSS projects I'm
associated with, jEdit (a great editor), has an java installer that does
a lot of neat things. The installer is also OSS, and could probably be repurposed.
I can find out where it can be downloaded from if you want.

Todd

-- 
 Todd Papaioannou @ Earth  http://www.luckyspin.org
 Luckyspin @ TerraFirmA   http://terrafirma.terra.mud.org
 Why search for life when you can create it? http://www.dalilab.com
.         .        .       .      .     .    .   .  .  . .

[Aglets-developer] Easy Installation of ASDK?

[Thomas C. <cal...@te...>]

Robert and I talked recently about making the ASDK easier to install for JDK
1.3. This is extremely useful for the technologists among us, but a major
PITA for a new user, especially a non-programmer who just wants to play with
mobile agents:

http://www.cis.ohio-state.edu/~muniandy/agletsd-j2/1_2/

I also know there are people and companies researching the ASDK for
commercial products. For those products to succeed, the ASDK needs an easy
and modular (modular in the sense of installing the ASDK plus the agents and
classes that are part of the custom application) installation.

At this point, we are thinking of a self-executable JAR file that asks for
an installation directory via a simple Swing interface. Once entered, the
installer extracts the ASDK from the executable JAR file to the indicated
directory.

On first run, Aglets goes into a quick Swing interview where the user
creates an account (certificate and key in the keystore connected to
username/password). On successive runs, the user enters username and
password as usual.

What do you think? Is anyone already working on this? Does anyone have code
snippets for installing from a executable JAR, creating a certificate and
adding it to the keystore, etc.?

Tom (Calivera)

[Aglets-developer] Aglets for JDK1.3

[<DT...@pi...>]

In reply to the following:

JDK1.3 compatible aglets can be found at:

http://www.cis.ohio-state.edu/~muniandy/agletsd-j2/

-------------------------------------------------------------

Date: Mon, 5 Mar 2001 10:18:54 +0100 (MET)
From: Sonia Perez Fuente <jib...@si...>
To: <agl...@li...>
Subject: [Aglets-users] Aglets and JDK 1.3


           Hi!!!!

           I'm new with aglets and I need some help, could somebody help
me??

           I have to run aglets with JDK 1.3 and I have read here that
somebody has do it yet. I would like to know where can I find aglets which
run under JDK 1.3 because I heve seen at
www.trl.ibm.com/aglets/dowload.htm that they can't run wiht JDK 1.2.


           Thank you for helping.


                               Sonia.


           PD: Sorry about my english ...

[Aglets-developer] Artificial Intelligence

[Daryl B. <Da...@sc...>]

Dear Robert (and fellow developers),
 
        Thanks for the link to the licence. I should have found that one
myself! I just thought it would be the same as the one sent with the SDK...
Doh.
        Anyway, remember my curiosity surrounding aglets and
genetic-programming/algorithms? [I've been on another project lately, so
that explains my absence from Aglets.] Well, my original idea for an
application of Aglets is to create a distributed problem-solving system
using an open-source genetic prog/alg system and aglets. The result would be
the type of system that is similar to what is rumoured to have helped crack
the human genome. Alternatively, if genetic programming is used, the result
would be a system that would help in developing algorithms to compute
solutions to problems -- for example, protien folding is supposed to be the
next big thing.... why not have everybody contribute to host not only
computing power, but also data sets that can be used to help evolve a
solution. Thus, solution aglets would become "local" to a range of test
cases, but then could move from computer to computer and cross-breed. One
computer, therefore, would not have to host all the tons of test cases
necessary to solve a problem. There could be a network of many computers
each hosting a range of test cases.
        This is much like what goes on in real life; people live in
different climates and conditions, but then shack up with people from other
places. Although human evolution is said to have ceased due to
over-population, and a lack of stress on human living conditions, the idea
still remains.
        When I came across Aglets, it lit a light in my mind. If a problem
was phrased in the open-source genetic algorithm language, it would be
possible to ask the "internet" (an aglets network) a question, and have it
eventually return an answer. It would also be possible for the agletnet to
evolve general solution algorithms, and thus contribute to human
understanding. What do you think? (that question goes to everybody on the
list!)
        I also have many other ideas on this subject that stem from my very
limited knowledge of biology coupled with my knowledge of computer science.
Unfortunately, I can't write them all down here right now.
 
Sincerely,
 
        Daryl.

RE: [Aglets-developer] Licence - What is IBM's interest in Aglets these days?

[Robert B. <rbergstrom@iSolutionsCorp.com>]

Daryl:

Good to hear from you again.

On August 23, 2000, IBM released the Aglets SDK under the IBM public
license.  You can find the annonuncement at
http://www.aglets.org/news/index.jsp which will also point you to the text
of the license.  There are sections which address commercial distribution.

Good luck with the project, :)

// RAB

> -----Original Message-----
> From: Daryl Beattie [mailto:Da...@sc...]
> Sent: Monday, March 05, 2001 10:24 AM
> To: agl...@li...
> Subject: RE: [Aglets-developer] Licence - What is IBM's interest in
> Aglets these days?
> 
> 
> Dear Developers,
> 
>         Hey, I have a question related to the last one 
> (licence agreement
> stuff), if you don't mind. It says in the licence that ships 
> with AgletSDK
> that the IBM licence expires on July 1, 2000. What licence is 
> the source
> under now? And where can I find a reference to that licence?
>         Also, sometimes it's not an important point, but what 
> are the laws
> regarding distributing aglets along with software that a 
> company sells?
> Sometimes that's not allowed, and the people a company sells 
> a program to
> have to download the third-party stuff themselves because of 
> licencing. Just
> wondering if there are any issues here.... Thanks. :) 
> Depending on what the
> issues are, I may or may not be able to start an aglets 
> project here at my
> company.

RE: [Aglets-developer] Licence - What is IBM's interest in Aglets these days?

[Daryl B. <Da...@sc...>]

Dear Developers,

        Hey, I have a question related to the last one (licence agreement
stuff), if you don't mind. It says in the licence that ships with AgletSDK
that the IBM licence expires on July 1, 2000. What licence is the source
under now? And where can I find a reference to that licence?
        Also, sometimes it's not an important point, but what are the laws
regarding distributing aglets along with software that a company sells?
Sometimes that's not allowed, and the people a company sells a program to
have to download the third-party stuff themselves because of licencing. Just
wondering if there are any issues here.... Thanks. :) Depending on what the
issues are, I may or may not be able to start an aglets project here at my
company.

Sincerely,

        Daryl.


 -----Original Message-----
From: Robert Bergstrom [mailto:rbergstrom@iSolutionsCorp.com]
Sent: Monday, March 05, 2001 10:12 AM
To: Aglets-Developer (E-mail)
Subject: RE: [Aglets-developer] What is IBM's interest in Aglets these day
s?


Dave: 
Q: What's IBM involvement with aglets since it's open source release? 
A: Some of the original authors lurk on the user list.  I'm not aware of any
development. 
Q: Who takes responsibility for the project? 
A: It an all volunteer effort.  I'm one of the admins of the SF project. 
Q: Is IBM doing any development independent of the open source project? 
A: None for which I'm aware. 
Q: Where does this leave developer who want to use Aglets commercially? 
A: IANAL, but the IBM public license seems to allow commercial use.  As far
as support is concerned, I don't know what IBM would offer.  There is this
list and the user list.  There are some people doing commercial research,
but nothing their willing to talk about. :)
Hope this helps, 
// RAB 

RE: [Aglets-developer] What is IBM's interest in Aglets these day s?

[Robert B. <rbergstrom@iSolutionsCorp.com>]

Dave:

Q: What's IBM involvement with aglets since it's open source release?
A: Some of the original authors lurk on the user list.  I'm not aware of any
development.

Q: Who takes responsibility for the project?
A: It an all volunteer effort.  I'm one of the admins of the SF project.

Q: Is IBM doing any development independent of the open source project?
A: None for which I'm aware.

Q: Where does this leave developer who want to use Aglets commercially?
A: IANAL, but the IBM public license seems to allow commercial use.  As far
as support is concerned, I don't know what IBM would offer.  There is this
list and the user list.  There are some people doing commercial research,
but nothing their willing to talk about. :)

Hope this helps,

// RAB

[Aglets-developer] What is IBM's interest in Aglets these days?

[<DT...@pi...>]

                                                                                                          
                                                                                                          
                                                                                                          
                                                                                                          
                                                                                                          
                                                                                                          
                                                                                                          
                                                                       Hi,                                
                                                                       I am new to this source forge      
                                                                       project. I have a general question 
                                                                       about this aglets project. If IBM  
                                                                       handed this over to open source    
                                                                       development, does this indicate a  
                                                                       lack of interest from mid-2000 on? 
                                                                                                          
                                                                       Who does take overall              
                                                                       responsibility for this project?   
                                                                       Is IBM developing independently    
                                                                       and with one eye on the open       
                                                                       source offerings? Where does this  
                                                                       leave developers who might want to 
                                                                       use this technology commercially?  
                                                                                                          
                                                                       Grateful for the answers to my     
                                                                       questions.                         
                                                                                                          
                                                                                                          
                                                                                                          


Dave Tyler

[Aglets-developer] [ aglets-Bugs-404714 ] Ticket scheme case sensitive

[nobody <no...@so...>]

Bugs #404714, was updated on 2001-02-27 13:55
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=107905&aid=404714&group_id=7905

Category: None
Group: None
Status: Open
Priority: 5
Submitted By: Robert Bergstrom
Assigned to: Nobody/Anonymous
Summary: Ticket scheme case sensitive

Initial Comment:
Thanks to Marc Mosko.

While playing with using HTTP messages to cause aglet 
dispatch, I found
that creating a ticket is case sensitive to the 
protocol name.  The
error
is really in java.net.URL.  See below.

Perhaps in Ticket.java, the function setDestination
(...) should have:

_protocol = destination.getProtocol().toLower();

Marc

----- Sample 1
** Request
farpoint.tear.com% lynx -dump
http://farpoint.tear.com\:4434/aglets/default/ddc/msg_D
ispatch\?
aglet=edu.ucsc.cse.http.mmosko.Sample\&dest=ATP%3A%2F%
2Froaster.tear.com%3A4434

** Tahiti console snips
Mon Feb 26 11:39:33 PST 2001 (BaseUtil): Message <
[Message : kind =
msg_Dispatch: arg =
{%querystring%
=aglet=edu.ucsc.cse.http.mmosko.Sample&dest=ATP%3A%2F%
2Froaster.tear.com%3A4434,
dest=ATP://roaster.tear.com:4434, cgi-response=,
aglet=edu.ucsc.cse.http.mmosko.Sample}: priority = 5]>

java.net.MalformedURLException: unknown protocol: ATP
        at java.net.URL.<init>(URL.java:290)
        at java.net.URL.<init>(URL.java:210)
        at java.net.URL.<init>(URL.java:229)
        at com.ibm.aglet.Ticket.getDestination
(Ticket.java:179)
        at com.ibm.aglets.LocalAgletRef.dispatch
(LocalAgletRef.java:881)
        at com.ibm.aglets.SystemMessage.handle
(SystemMessage.java:200)
        at com.ibm.aglets.AgletThread.run
(AgletThread.java:106)


------ Sample 2
farpoint.tear.com% lynx -dump
http://farpoint.tear.com\:4434/aglets/default/ddc/msg_D
ispatch\?
aglet=edu.ucsc.cse.http.mmosko.Sample\&dest=atp%3A%2F%
2Froaster.tear.com%3A4434

(works fine)


----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=107905&aid=404714&group_id=7905

[Aglets-developer] [ aglets-Bugs-404709 ] Improper parsing of CGI query string

[nobody <no...@so...>]

Bugs #404709, was updated on 2001-02-27 13:44
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=107905&aid=404709&group_id=7905

Category: None
Group: None
Status: Open
Priority: 5
Submitted By: Robert Bergstrom
Assigned to: Nobody/Anonymous
Summary:  Improper parsing of CGI query string

Initial Comment:
Thanks to Marc Mosko.

The current (1.2) parser in HttpCGIRequestImpl.java 
does not 
handle "/" characters in a CGI query string.  It 
considers them token
separators (like the &).  The token delimiter should 
be removed when
the parsing gets to the query string so the "cgi" 
variable will be set
to whatever is remaining in the line.

It would also be nice to be able to pass relative path 
information
to an aglet (e.g. 
http://aglet:4434/default/aglet/mine/relative/path), 
but that's a bigger change.

Marc

farpoint.tear.com% diff HttpCGIRequestImpl.java 
HttpCGIRequestImpl.orig
264,265c264
<                       // Marc M - change delimiter 
to "" to get to EOL
<                       cgi = t.nextToken("");
---
>                       cgi = t.nextToken();


----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=107905&aid=404709&group_id=7905

[Aglets-developer] FW: Aglets-developer -- confirmation of subscription -- request 456436

[Thomas C. <ald...@ph...>]

-----Original Message-----
From: agl...@li...
[mailto:agl...@li...]On Behalf Of
agl...@li...
Sent: Monday, February 26, 2001 3:51 PM
To: ald...@ph...
Subject: Aglets-developer -- confirmation of subscription -- request
456436


Aglets-developer -- confirmation of subscription -- request 456436

We have received a request from 64.128.168.221 for subscription of
your email address, <ald...@ph...>, to the
agl...@li... mailing list.  To confirm the
request, please send a message to
agl...@li..., and either:

- maintain the subject line as is (the reply's additional "Re:" is
ok),

- or include the following line - and only the following line - in the
message body: 

confirm 456436

(Simply sending a 'reply' to this message should work from most email
interfaces, since that usually leaves the subject line in the right
form.)

If you do not wish to subscribe to this list, please simply disregard
this message.  Send questions to
agl...@li....