I have given up, been trying to work out how to get this thing to work, when I have the patience to do so, as per http://www.cakewalk.com/Support/Knowledge-Base/2007013069/Using-the-Novation-ReMOTE-SL-as-a-controller-with-SONAR#Chapter4
 
I can get the sliders to work the volume on the first 8 tracks but have not been able to get them to bank switch through hardware, same with the bottom row of knobs for pan, had some tempory success with play, stop and rewind on transport, and some things simply will not let me use midi learn.  I got lots of buttons, with the CC paramters setup via the guide in HW, surely there must be someway to get mute and solo to work at least?
 
I think I am either doing something really stupidly wrong, or maybe these device just do not like sonar/windows

AFAIC, the Automap utility relies on HUI protocol, which is not supported in Sonar OOTB. Try AZ Controller http://forum.cakewalk.com/Sonar-Resources-and-Utilities-m3392713.aspx
It worked for me. Although I use M-Audio Code keyboard, the basic problem with HUI protocol support in Sonar is the same.

The automap sware/virus has been deleted by me from my computer, well as much of it as I could.  Think there are still some files it could not uninstall itself.  I do not want to be dependent on sware updates from a tertiary vendor to keep a controller working or my audio files editable.
 
I am using what they call the advanced mode on SL Mk2, as a generic type midi controller, where you manually assign the values and descriptors by writing to eproms on the midi controller hardware, and trying to get it to work with ACT template within sonar platinum as per the described method in the cakewalk knowledge base.
 
It works fine as a Cakewalk Generic Surface.  Sends, arm, automation write, mute, solo, transport, volume, pan and bank changing work in this mode.  So Sonar must understand or be capable of understanding what the controller is sending.
 
To my understanding the only real difference in what the SL Mk2 device sends is that some midi values are set to 0 instead of 64 for some continuous controller numbers.
 
Where it all goes semi peared shaped is in midi learn, the pan on the bottom encoders is ok, the sliders for volume are ok, and tonight I finally got basic transport to work, but the rest is ******.
 
Maybe I need to study more about the different protocols, but seriously, its just a midi controller, I should not need to be this difficult to get a midi controller to work with a bit of software.  This is 2016 with windows 10 and plug and play, not 1986 with PC-DOS 3.x and windows 1.0 when you had to configure everything from tables in several instruction manuals.

Siluroo
Maybe I need to study more about the different protocols, but seriously, its just a midi controller, I should not need to be this difficult to get a midi controller to work with a bit of software.  This is 2016 with windows 10 and plug and play, not 1986 with PC-DOS 3.x and windows 1.0 when you had to configure everything from tables in several instruction manuals.

MIDI is still the same since 1986 Piano was not significantly changed in mechanics since 100 years and still no one can play it professionally in 15 minutes without spending years to learn how to play it.
 
Sonar "MIDI Learn" works with "simple" controls only (not endless). "ACT MIDI" / "Generic Surface" / (my) "AZ Controller" are the way to use any advanced controls. Note that you still need to read the manual (chapters 20 and 40 in the Sonar reference guide for the first two, my manual for the last) to get some controller do what you want. Computers are still not good in reading your mind...

I have downloaded the X3 version reference guide, in my quick search I did not find a Platinum version in pdf, just the online version, online is great if you already know at the answers or at least the right terms to search for, but not so great for browsing through in a speed reading session.
I am not trying to do anything that I would call advanced, but to me advanced is getting into macro or programmable function territory.
Will have another bight at this later in the week when my mind is more fresh, and its not 1.40am in the morning.
Personally, I think its time for a new version of midi, though I am not sure if what is being discussed by the MMA is what we need, as the impression I get is that its being designed for large scale high end systems with thousands of channels (http://www.synthtopia.com/content/2015/01/16/new-midi-hd-protocol-has-reached-a-milestone/), thats if the members of the association do eventually agree. 

Siluroo
I am not trying to do anything that I would call advanced, but to me advanced is getting into macro or programmable function territory.

Let me explain the situation is several sentences...
 
Each parameter (in Sonar, VST) has a value from 0.0 to 1.0. F.e. centered pan is 0.5 (strictly 1/2).
"Normal" MIDI message has values from 0 to 127, some from 0 to 16535. That is what simple controls, like faders and knobs are sending. So, assuming 0 is 0.0 and 127 (16535) is 1.0, the logic is strait forward. That what is working for "MIDI Learn" in VST and "Remote control..." in Sonar.
 
Well... there is already a problem. Like there is no "perfect tune" in modern notes (12 devisions of octave is a "compromise"), there is no perfect 1/2 from 127. 64 is "a bit less", 65 is "a bit more". That is why without future "tricks" it is impossible to set perfect center using strait approach (most software take that into account, some not).
 
With endless encoders, the meaning of "turning" is just +1 or -1 (sometimes +2,+3,etc. to specify the speed). But which interpretation of "+1" should be used? For some situations 1/128 is too coarse (f.e. EQ Frequency), for some 1/16535 is too fine (you will need to turn the encoders many full rounds to change volume by 1dB). And so we slowly move toward some "macro or scripting" territory.
 
But the number of parameter in modern DAW, including all all parameters from all VSTs fast explode toward 10000 and more even in a small project. And you have 20-50 hardware controls only on surface. And so, without scripting, even with "modifier keys" it is impossible to cover all of them.
 
Which scripting is "perfect"? There are several "standard" proposals how limited number of hardware controls should operate endless number of DAW parameters. Most famous is Mackie Control. But there are other, like AutoMap, NKS, etc. For different purpose, with different approaches. All of them have something common, all of them are not "perfect" for each particular user.
 
And so different DAW have different solutions to allow end users decide what is the best for them. With general groups:
* simple - MIDI Learn particular hardware parameter for particular software parameter, no scripting but the number of controlled parameters and they way they are controlled are fixed
* "standard" - complicated scripted but fixed mapping, made by "wise" developers for particular devices (in case of Sonar, specialized plug-ins)
* "advanced" - generic mapping, with some freedom in parameters and hardware definitions, but still more or less fixed "scripting" logic ("ACT MIDI", "Generic Surface")
* "expert" - arbitrary user defined logic (Cakewalk Control Surface SDK for C++, AZ Controller with own philosophy and IDE)
 
What I want to say is that you can get simple functionality without big effort, but when you want more you need at least something from "macro or scripting" world.
 

That is a very well written explanation of the mess the industry has got itself into, and not out of!
There have been midi controllers for 30+ years, the problem of changing a value in any DAW should have been solved by now to a level where a half blind 90 year old with dementia after a full bottle of bourbon could still assign a generic controller endless rotary encoder from any controller to any controllable parameter.
To put things into the context of which I see things, 30 years ago most cars did not have a computer in them, at least not of any real significance, unlike DAW's, which are a computer, however yesterday I read in one of our national newspapers that Ford will be offering a fully automated driverless car for sale in 2021!  So in 2021 it is most likely that the same half blind 90 year old with dementia after a full bottle of bourban will be able to go on a national solo pub crawl in their own car as long as they can still keep slurring out some basic directions, but not be able to their midi controller to work! I will also bet that the driverless car from Ford does not need to reconfigured each time it changes which road it is on!
 
DAW's are now at a normal consumers price point, they are not just in the domain of professional studios which have dedicated sound technicians, its mums and dads, 10 year olds, and grandpa and grandma that are the operators.  They are mass market products now, the midi controllers are mostly retail at less than what it cost to take the family out for a meal.
 
What I am trying to say is that its ridiculous that we are still configuring devices like it is 1986 still.
 
Btw, I do appreciate the help and explanations given, and I am not being at all critical of any one organisation or person, I am basically amused at how backward things are with regards to the ease controller and DAW compatibility in general.
 
I also was hoping to not have to relearn midi or study any other protocol to this level, but I will go read the technical manuals and web articles and work it all out probably on the weekend when I have a heap of uninterrupted time to focus on it.  Luckily or unluckily for me, I have been configuring computer devices for over 30 years, have some programming skills, which make me technically literate, unlike most of the worlds population.
 
It is annoying, and gets in the way of doing what I want to be doing.
 
On the brights side, I should learn a little more about sonar in the process.
 
Now back to downloading Wusik V8.02 and associated libraries, in between short sleeps.

Novation has made AutoMap for that. As long as you use it, you are not forced to deal with MIDI and scripting. That software has some disadvantages. In your comparison with Fort, let say your car takes 5x longer route than required or refuse to build the route with the message "selected destination can not be reached" (happens with my navigation). So that was your decision to turn "the computer" off.
 
Note that once everything is configured right (either AutoMap or ACT way), you can assign a knob to parameter rather quick.
 

 When I had problems with configuring ACT mode on the Novation SL, I installed automap, which I was loathe to do, since the way it works with VST's, but yes, you can map things with it, but when it kept periodically crashing on me, and exhibiting some behavior symptoms that reminded me of memory leaks in days past, I uninstalled it.
 
Computer navigation is only as good as the data subset you are subscribed to, and no matter which system you currently get is still full of errors, even google map is not that accurate, for instance the address at which I live is shown incorrectly on google, our neighbors house block overlay covers our houses aerial photo and the main road is also offset.  To understand why this is you need to be familiar with the nightmare that is cartography, which has multiple different mapping standards, projections, and distortion technique used to to cobble together semi accurate 2d model of 3d object.
If you have an curious type of mind, you might find the next two links interesting reading
https://en.wikipedia.org/wiki/Map_projection
http://www.theatlantic.com/technology/archive/2012/09/how-google-builds-its-maps-and-what-it-means-for-the-future-of-everything/261913/
 
Basically getting a value from an encoder of some sort is a hugely easier task than getting accurate location data mapped.  For with an encoder you have a known good value, with maps you start with an approximated value from a an already distorted and inaccurate model.
 
You would be amazed at then number of land ownership disputes that have been triggered by more accurate mapping technology in last 20 years.
 
 

Siluroo
Basically getting a value from an encoder of some sort is a hugely easier task than getting accurate location data mapped.  For with an encoder you have a known good value, with maps you start with an approximated value from a an already distorted and inaccurate model.

Getting the value from an encoder is not a problem. The question is what you want to do with this value. That can be use case and user specific. Not only "fine" and "coarse" setting, but also behavior in case of several discrete settings (f.e. EQ type, I prefer one tick switch the type while other prefer some "range" does not change it). But real challenge is the definition which parameter one encoder controls under different conditions. 
 

You would be amazed at then number of land ownership disputes that have been triggered by more accurate mapping technology in last 20 years.

My wife has studied geology, in particular processing of corresponding maps. Also my uni mates was processing geographic maps collected over time (starting from hand drawn some centuries ago till satellite photos). So I am a bit aware that is not an easy job