directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alex Karasulu <aok...@bellsouth.net>
Subject Re: Re: [asn.1] Decoder service design ( was: Paper on original SNACC)
Date Fri, 23 Jan 2004 19:44:11 GMT

> From: "Wes McKean" <wmckean@logictrends.com>
>
> Well, the time spent **blocking** would be negligible, since we are talking 
> about client/server network communications.  We can expect the entire 
> message to arrive in a reasonable amount of time.  Most messages will arrive 
> inside one byte buffer anyways.

Depends on the PDU size but I pretty much agree with you on most of what
you're saying.

> Continuing our discussion...
> 
> Yes, so I think I have the general idea now.  The thread doing the reading 
> reads in these byte buffers and looks up our decoder.  The decoder doesn't
> do anything but break the incoming message down into a tree of TLV values. 
> This is MUCH easier to do, since all the decoder is bascially dealing with 
> at this point is TLVs, not ASN.1 messages.  Once the tree is complete, then 
> we pass it off to something else to interpret the tree and send the message 
> off to be processed.

Sounds good to me, could you carve up some documented interfaces.  I could
then work with you to build the SEDA Stage for your decoder.  Looks to me
you're going to implement a 

> 
> This gets me excited, cause it is elegant and is totally different from the 
> way the current, stream/message based decoder works.

Oh dude this is getting really cool.  I think the whole interpretation
of the TLV tree will be really interesting now and probably all that much
easier and cleaner to implement.  

We could also do some interesting things to make the server faster by not 
converting certain ASN.1 datatypes into Java types for comparison ops.  

We would need to talk about this though.  I think you're opening the door to 
something new for us and we will not know about its full potential until 
we start seeing the data inconversions happening in the V part of the TLV.

> So, if we are in agreement, then this is the approach I will take...

Oh yeah.  But I wanna work with ya too or at least look on.  You got me
pumped up man.  Let's talk interfaces so we can divy up the task of fitting
this thing into the server.  Cuz you may not want to deal with putting it
into the Merlin and SEDA stuff.  I can focus on that and you can focus 
strickly on the ASN.1 stuff.  Now that we're making our components POJOs
with container wrappers for using in multiple containers this should be 
cake.  We got a nice separation of concerns here.

Let me know what interfaces you intend to expose.  I'll then start building
them into the server that way the integration is ready once you have the 
decoder working.

While doing this we still need to focus on documentation too.  I've just
had so many things going on lately, both work and health related.  
Having walking pneumonia (can't spell that) has screwed me over lately 
with the stupid doctors visits.  The inlaws are visiting and I have to
clean up my computer hole.  So you know how it is ... 

Anyway let's rock as best as we can on both the site and this little decoder
component.  It's a good exercise for the both of us.

Till l8r,
Alex



Mime
View raw message