Newton Toolkit
Volume Number: | | 9
|
Issue Number: | | 11
|
Column Tag: | | Newton
|
|
Newton Toolkit
Youve seen Newt and what it can do - now take a look at the tools developers will work with.
By Don Bresee and Neil Ticktin, MacTech Magazine Editorial Staff
Tony Espinoza, Newton Software Tools, Apple Computer, Inc.
By now, youve taken a look at Newton. (If not, you should go and read the editorial starting on page 4 of this issue.) But as developers, you want to know about the tools for Newton - specifically the development environment. The Newton Toolkit (NTK for short) is how you develop apps for Newton. Below is a description of Newtons built in tools, NTK and NewtonScript.
The Newton Toolkit
First, you need to know some basics about Newton programming and about NTK itself.
Newton Basics
The Newton system is an extremely versatile object oriented environment designed around the concept of the view. A view is any graphic component of the Newton user interface. Views also serve as the basic building blocks of Newton applications. You design and build these views with the Newton Toolkit for the Mac, which we will talk about later. First, it is necessary to get aquainted with the NewtonScript programming language.
NewtonScript
NewtonScript is a full featured, object oriented programming language. It has all the programming constructs and data structures needed to produce sophisticated applications. Some highlights:
All statements are expressions (they return a value),
Statements are grouped with Begin End,
Loop constructs For, While, and Repeat Until are provided,
A Foreach statement allows looping through the elements of a data structure,
Break statement for terminating loops,
Functions and Methods with parameters and local variables,
Return statement for functions and methods,
Arithmetic and string operators,
Class and parent / child inheritance.
The most basic data reference in the Newton system the long word (32 bits). A reference is either a pointer or an immediate reference. Immediate references can be integer, character, or boolean and use 30 bits for data and 2 bits for flags. Booleans have two possible values, TRUE or NIL. Characters are represented in Unicode (in two bytes) to facilitate international conversions. Pointers reference objects in the heap. These objects include:
Symbols (variable names, field names, script names),
Strings,
Reals (64 bit floating point numbers),
Arrays (numbered elements, starting with 0),
Frames (named elements).
Symbols are useful in writing general code. For instance, you could write a function that takes a symbol parameter that is the name of another function to call in a certain situation. The same idea can be used for field names in data structures. A routine could take a field name as a parameter, and perform some operation on that field, allowing you to write more general code.
The array data structure is extremely useful. The elements can be of different types in the same array. Elements can be added, deleted, and sorted. This type is ideal for lists of objects. Arrays can be given a class type. This seems to be for readability reasons since the class does not give an array any special properties, just a name. This is different than frames, which we will talk about now.
The frame is the most important type in the Newton system. A frame consists of a collection of slots. Each slot has a name and a value. A slot value can be a function, making the frame type capable of class definitions. Frames have special slots for class names and links to other classes that are needed for the inheritance mechanisms.
There are two types of inheritance in NewtonScript. The normal class inheritance that we are used to is called prototype inheritance in NewtonScript. When a frame is defined, the superclass name should be put in the _proto slot. Then the new frame will have access to (inherit) all the slots of that class, this includes data and methods.
The second type of inheritance is called parent inheritance. This mechanism is similar to the chain of command used in the THINK Class Library (TCL). This type of inheritance is used to set up hierarchies (views that contain and manage subviews). When a frame is defined, the class name of the parent (or supervisor) should be put in the _parent slot.
These two inheritance mechanisms interact in much the same way as they do in TCL. When a message is sent to an object, that object is checked first. If it can handle the message, it does. If not, the prototype (superclass) is checked, and then its prototype. If this process bottoms out (no prototype of the initial object can handle the message), the message is passed along to the parent (supervisor). Then the same process is used for the parent object (if the parent object cant handle the message, the prototypes are checked and then that objects parent, and so on). For example, in a TCL application, when a quit message is sent to the gopher (assume it is a view), the gopher and its superclasses attempt to handle the message and cannot. The message is passed up the chain of command to the window, then to the document, and then to the application which completes the handling of the message (probably by a superclass of your application object).
The Newton System
The view is the most important object in Newton applications. Views are created from templates. A template is a frame containing the data description of an object (a view). The relationship between views and templates is similar to that of objects and classes.
Dozens of proto templates are provided in the Newton ROM. These templates offer most of the user interface functionality you will need when creating applications. It is quite complete. It includes buttons, pictures, text (static and editable), geometric shapes, pick-lists, and plenty of others. In addition to a view classes specific messages it handles, there are several system messages that can be sent to any view. These messages will allow you to customize your views behavior by supplementing the normal system behavior when certain events occur. For instance, you may want a certain view class to do something every time it is drawn. You simply define a viewDrawScript method for your view class and it will be called each time the system draws a view of that type.
Newtons data storage and retrieval model is very interesting. It uses a unified data model, all data is stored in a common format. Therefore, no translation is required to read another applications data. The model is built around soups. A soup holds a collection of related data objects called entries. Each entry is a frame. There is no restriction on the format of the entries. Different entries in the same soup can have different slots. Of course, this can get out of hand very quickly and it is a good idea to keep the structure of the entries uniform within each soup. Optional slots are useful for minimizing the number of bytes needed for an entry. This is an important point because a soup is a persistent storage object and memory is precious. In fact, when entries are saved to or retrieved from a soup, they are compressed or decompressed behind the scenes. This is done to make best use of memory. You can also define an index for each slot in a soup that will be searched frequently.
Each soup has a name and is contained in a store, which is also named. A store can be thought of as a disk volume. The Newton contains one internal store. Additional stores can be present on PCMCIA cards and other devices. To retrieve entries from a soup, you call the Query function. This function returns a cursor object. A cursor accesses the set of entries that satisfy the query, and gives the ability to step through those entries in indexed order. The system provides functions to do whatever you might want with stores, soups, entries, and cursors.
Another important concept on the Newton is Routing, which simplifies communications. Print, Fax, and Mail are all handled through one API. You define a format (a view) for the data and specify that format in the routing frame. You also design a routing slip that will get any additional information needed from the user before the operation is commenced (such as a name and fax number to send a fax). The system takes care of the other details for you.
Newton Toolkit
The Newton Toolkit (NTK) is the environment provided for the development of Newton applications. The main part of NTK is a Mac application that allows you to define views and write scripts. It also allows you to combine layout (templates), resource, data, and driver files into a project. NTK does this with a project document in much the same way THINK C does. The project window lists all the files in the project. Double clicking on a file in the list opens it for modification. Building your project results in what is called a package.
Apple recommends that this application be run on a Quadra with 8 MB of RAM and System 7.1. I played with it on a IIci and it took one minute fifteen seconds to launch from the Finder, and one minute ten seconds to open the checkbook sample. Once a project is open, it runs plenty fast enough. Expect installation of NTK to use about 2.5 MB of disk space. [We are reviewing 1.0b4 version. 1.0b6 is reportedly five times faster and will therefore easily run on a 68030 machine or better. Expect the release version to have better performance. - Ed.]
NTK also provides a Newton application called Toolkit which allows you to download your package to the Newton and test it (accessible on the Newton through the Extras drawer). You can test it and see the results immediately. This allows you to develop your application incrementally. You work on a view, build a package, download the package to the Newton, and then test it. You do this routine until that view is working right and then you build the next view.
Views and templates are created using the Template Editor. This editor has a layout window for drawing views (which is done by dragging out rectangles like in you would any drawing application), and a palette for selecting the view class. This palette has all the predefined Newton interface elements (templates), as well as any templates you have defined in the current project. Items are added by selecting an item from the palette and then placing the view where you want it in the layout window. A preview menu item allows you to view the screen as it would look on the Newton.
Fig. 1 Template Editor, palette, and Project Window.
The Browser window allows you to modify those templates created in the Template Editor. This is where you write the views specific methods. It also provides you with the ability to edit the initial values of all slots in the template. When you open a Browser, the view selected in the Template Editor and all of its subviews will be available in the new window. You can easily jump from method to method with pull down menus or with a pick list.
Fig. 2 Browser Window
The development process is slick considering the fact that you are developing applications for one computer on another. First you need to connect the Newton and the Mac. You can do this with a null modem cable (through the serial port) or with a LocalTalk connection (through the printer port). [With the pre-release version you want to stick to serial only. LocalTalk has not been tested in 1.0b4 and has problems. - Ed.] When the two are connected, you tell NTK to install the Toolkit application onto the Newton. Then you open the Newtons Connection tool in the Extras drawer and tell it to connect to the Mac. After all this, the Toolkit tool will appear in the Extras drawer on the Newton. Now we are ready to download your project. Open your project and build a package in NTK. Then you tell NTK to download the package. Open the Toolkit tool on the Newton and hit the download button (you will need to choose a Mac if you are using a LocalTalk connection). Your application will appear in the Newtons Extras drawer. Open it and test, test, test.
This sounds like a long process, but most of it will be done once per development session. When the Newton and the Mac are connected, and the Toolkit is installed on the Newton, only the build package and download steps are needed. The new copy of your application replaces the old one. So testing is quite easy after everything is installed and connected.
Some Examples of Newton Code
So now youve seen some of NTK, but what does NewtonScript really look like? To demonstrate this, well look at some examples of programming using NTK, NewtonScript and Newton itself.
Adding to the Intelligent Assistant
You probably know a bit about Newtons Intelligent Assistant. This built-in tool is designed to help the user use Newton in a human fashion. For example, if you are in the notetaking area, write in Lunch with Neil Tue at 2pm, Newton will take it from there - looking up Neils information in the name cards and scheduling the lunch in your calendar.
But what if you want to add verbs to the list that Newton knows how to handle. For example, if you were to ask it to add a verb called page that would have Newton page the person you asked it to. You will have the option of a paging service through which you would like to page.
To take advantage of this or check it out, first you need to have downloaded NTK Toolkit to Newton and connect/opened the Inspector window. Then, type in each of the frames below separately (the actual syntax, dont bother with the comments which always start with //) into the Inspector window. Once you have it in, select the text, press enter to evaluate each one (remember to select the whole frame before hitting enter [similar to MPW]). You should get back a different response to each frame processed.
// 1
Inspector Window Code:
// the action template describes the "verb".
// requires a slot named "isa" that has 'dyna_user_action
// requires a value slot, which should be a string (never used)
// requires a lexicon slot, which should be a word array (synonyms)
actionTemplate := {
value: "foo",
isa: 'dyna_user_action,
lexicon: ["page", "beep", "zap"],
}
// the target template describes "nouns" for the task.
// requires an isa slot, which has the value 'dyna_user_obj
// requires a value slot, which should be a string (never used)
// requires a lexicon slot, which should be a word array (synonyms)
targetTemplate := {
isa: 'dyna_user_obj,
value: "bar",
lexicon: ["mobilecomm", "skytel", "pactel", "apple"],
}
// the post-parse function is called when IA matches the action.
// use compile function because it's a closure and it gets cloned,
// if just wrote a func() body, the clone would be huge in inspector.
// Don't need to use compile if it's written as part of an NTK app.
ppf := compile("print(\"Your Code Here!\")")
// the task template binds it all together.
// requires an isa slot, which should be 'task_template
// requires a primaryAct slot, which holds an actionTemplate.
// requires a signature slot, which is an array of actionTemplates,
// taskTemplates, and/or system task symbols. (i.e. 'person)
// requires a preconditions slot, which matches signature array, but
// contains symbols, one for each template.
// requires a postParse slot, which is the function to call
// requires a taskslip slot, which is a view template that can be
// opened by the postParse method to provide more UI. (not used)
taskTemplate := {
isa: 'task_template,
primary_act: actionTemplate,
preconditions: ['pagingNet, 'pageAction, 'whoToPage],
signature: [targetTemplate, actionTemplate, 'person],
postparse: ppf,
taskslip: nil,
}
// use global function RegTaskTemplate to give template to IA.
// keep the result, will need it later to unregister task (see below)
result := regtasktemplate(taskTemplate)
// use the parseUtter command to test.
p := parseutter("page Bob via MobileComm")
// once the above line is processed, the postParse method is called
// you will then see:
"Your Code Here!"
// here's the result frame which is also "self" within postParse func
// it's a clone of the taskTemplate passed with some additnl slots.
// parse, input, and raw contain frames based on templates matched.
// you can loop through them and use isa() to test for which of
// your templates were matched (see Checkbook example)
// phrases contains the matched words.
// noisewords contains unmatched words.
// origphrase contains the original phrase as written to IA.
// pagingNet, pageAction, and whoToPage are added because all the
// templates in the signature array were matched. The slot
// contains the result of the match for the given template.
#4415C19 {isa: task_template,
primary_act: {value: "foo",
isa: dyna_user_action,
Lexicon: [#4412121]},
preconditions: [pagingNet, pageAction, whoToPage],
signature: [{#4412241}, {#4412041}, person],
postParse: <CodeBlock, 0 args #44124D1>,
taskslip: NIL,
parse: [[#4415C69], [#4415C79], [#4415CA1]],
input: [{#4412041}, person, {#4412241}],
raw: [[#44128B9], [#44151A1], [#44127E1]],
phrases: ["page", "Bob", "MobileComm"],
noisewords: ["via"],
OrigPhrase: ["page", "Bob", "via", "MobileComm"],
pagingNet: [[#44127E1]],
pageAction: [[#44128B9]],
whoToPage: [[#44151A1]]}
// Once you are done with this command, you need to unregister the
// task to remove it. Remember to do this before re-registering the
// same task!
unregtasktemplate(result)
After you register the task (and until you remove it), you will notice that the Page verb is in the Assistant verb list just as all of the other verbs are.
Making your application screen-size independent
Just like the Macintosh, you will want to make your Newton applications screen size independent. The function call GetAppParams() returns a frame describing the current device view bounds and the location of the buttons. The return frame looks like this:
// 2
{appAreaTop: 0,
appAreaLeft: 0,
appAreaWidth: 240,
appAreaHeight: 320,
buttonBarPosition: 'bottom} // one of 'top, 'bottom, 'left, 'right
You can use this call to set the viewbounds of your base application view during at viewSetupFormScript time. The simplest way is to create a function that will resize your view for you and call it from your viewSetupFormScript. Here is one that will take up the entire screen (leaving space for a frame around the base view):
// 3
sizeBaseToDisplay
func()
begin
// assumes 1 pixel frame & top left parent relative justification
local b := getAppParams();
self.viewbounds.top := b.appAreaTop + 2 ;
self.viewbounds.left := b.appAreaLeft + 2;
self.viewBounds.bottom := self.viewbounds.top + b.appAreaHeight -
4;
self.viewBounds.right := self.viewbounds.left + b.appAreaWidth -
4;
end
Once your base application view is sized correctly, you can use the viewJustify slot to make other parts of your application parent releative or sibling relative justified. Not all parts of your application need to be full or center justified, some parts will not be affected by a slightly smaller screen size.
Other parts of your application may need to be dynamically constructed (a view that has children representing soup entries, like the checkbook example included with NTK).
If you follow these simple design and develop guidelines, your application should work on future Newton products.
A Serial Connection Sample
This simple sample shows how to make a serial connection with a simple protocol implemented. In order to test this app you need to have the Newton connected to an outside source, such as a computer with a terminal emulator running (i.e., a Macintosh running MicroPhone), and you will control the protocol by typing commands in the terminal window and Newton will respond. Of course this is easier if the control is inside an appplication.
To take a look at this, read through the below user instructions to get an idea of whats going on. Then build the below package in NTK using both the graphical tools and typing in scripts as indicated. Download this package on Newton. Then return to the User Instructions to run the example.
User Instructions:
1. Open up the terminal emulator (make sure local echo is on). The terminal emulator should be set for 9600bps, 8 bit no parity, XON/XOFF.
2. Start the Newton app
3. Hit Connect on the Newton side (now Newton will start waiting for an incoming ACK)
4. Type ACK? on the terminal, you should immediately get back ACK!
5. Type various other commands, and you should get input from the Newton.
(The following entries and responses are tied to entries in the Names card file on a specific Newton. Use names that you know are in your phone book.) Heres an example of a simple session:
ACK?ACK!
CARD!Walthrop,Royce,Hillsdale
Thomas,Linda,Newton
Anderson,Bob,Fine
NAME!Jill UK (You have to have something stored in the name preferences!)
The Protocol:
So what can you do with this? Newton will passively wait for a response from the other end. The other end should send a string with the four characters ACK? Newton responds with the string ACK! After this you could send the following strings to Newton and it will respond:
CARD! -> Newton will return parts of the card soup, strings separate
by CR+LF
NAME! -> Newton will return the owner name as a string
SIZE! -> Newton will return available RAM size in bytes (string)
SEND! followed by a string terminated with a >, as in SEND!this is
a string>
BYE! -> Disconnect, and the Newton end will signal that the other end
is no longer active (this is not fully support, you have to close the
app in order to disconnect).
The Code:
Create a project called "Serial Protocol" and create the following views/frames in NTK.
This first frame is created by the template editor in NTK. Dont type it in, just use the information below to create the view in the browser.
// 4
baseView :=
{_proto: protoapp,
title: "Dump Names",
viewBounds: {left: 0, top: 0, right: 240, bottom: 336},
};
This second frame also isnt code to type in, just the params you need to set up in NTK browser:
//5
info :=
{_proto: protostatictext,
text: "This simple application will dump all the Names Soup Information
via the serial port",
viewBounds: {left: 10, top: 18, right: 226, bottom: 66},
viewJustify: 0,
});
The DumpButton frame is partially created with graphical NTK tools, but much of the below scripting is to be typed in.
// 6
DumpButton :=
{_proto: prototextbutton,
text: "Connect",
buttonClickScript:
// Make the initial connection, start listening to incoming msg
func()
begin
// do a connect
anErr := ep:Connect(nil,nil);
if anErr then
begin
Notify(kNotifyAlert, "DumpNames",
"Check your cables and terminal emulator, couldn't connect");
return;
end;
// set the next input spec for listening mode
ep:SetInputSpec(ep.waitforACK);
SetValue(result,'text, "Connected, waiting for ACK?");
end;,
viewBounds: {left: 50, top: 218, right: 154, bottom: 250},
viewSetupDoneScript:
// Create the endpoint that we need later.
func()
begin
local anErr := nil;
// create an endpoint for our use
ep := {_proto: protoSerialProtocol, _parent: self};
// we assume 9600bps 8 bit no parity is fine
// instantiate the endpoint
anErr := ep:Instantiate(ep, nil);
// configure (XON/XOFF support)
ep:SetOptions([myInputOptions,myInputOptions]);
if anErr then
begin
Notify(kNotifyAlert, "DumpNames",
"Sorry, serial port already in use!");
return;
end;
print("Created the endpoint");
SetValue(result, 'text, "Ready for Connection");
end,
ep: nil,
viewQuitScript:
func() // release the endpoint (connection, memory)
begin
ep:FlushInput ();
ep:FlushOutput ();
ep:SetInputSpec (nil);
ep:Abort();
AddDelayedAction(func() begin
// let the endpoint clean up from the abort before
//disconnecting and disposing
ep:Disconnect();
ep:Dispose();
end,
[], 1000);
print("Disposed the endpoint");
end,
DumpResult: nil,
ReleaseEndpoint:
func() // release the endpoint (connection, memory)
begin
ep:FlushInput ();
ep:FlushOutput ();
ep:SetInputSpec (nil);
ep:Abort();
ep:Release();
ep:Dispose();
print("Disposed the endpoint");
end,
myOutputOptions:
{
label: kCMOOutputFlowControlParms,
type: 'option,
data: {
xonChar: unicodeDC1,
xoffChar: unicodeDC3,
useSoftFlowControl: true,
useHardFlowControl: nil
},
},
myInputOptions:
{
label: kCMOInputFlowControlParms,
type: 'option,
data: {
xonChar: unicodeDC1,
xoffChar: unicodeDC3,
useSoftFlowControl: true,
useHardFlowControl: nil
},
},
protoSerialProtocol:
// this is our special endpoint with the protocol built in.
// This works fine as long as you remember that this entity is
// created in Read-only space. Stay tuned for a solution where
// this is created in RAM (actually move this code so it
// executes during viewSetupDoneScript time and that's it)
{
_proto: protoSerialEndpoint, // the basic serial endpoint
// the rest here is just code for the state machine, all the
// state frames are stored inside the the endpoint itself, it's
// just a handy place to put them, those could be stored
// anywhere.
// We need to catch the last abort exception, later this should
// be more uniform, check out what the exception was and ignore
// the abort one (the one causing the notify)
exceptionHandler: func(exception)
begin
print("Got the exception from Abort, ignoring it!");
end,
// this is just the initial handshaking part to make sure that
// both ends are alive
waitforACK:
{
InputForm: 'string,
endCharacter: $?, // ACK? expected
InputScript: func(endpoint, s)
begin
if (StrPos(s, "ACK?", 0)) then
begin
// tell status if needed
endpoint:Output("ACK!", nil); // send response
endpoint:FlushOutput();
// the main dispatch loop
endpoint:SetInputSpec(endpoint.waitForFUNCTION);
end
end,
discardAfter: 200,
},
// This is the generic dispatcher state, send something ending
// with ! and the Newton will serve.
waitForFUNCTION:
{
InputForm: 'string,
endCharacter: $!, // expects a '!' as part of command
InputScript: func(endpoint, s)
begin
if(StrPos(s, "CARD!", 0)) then // card function
begin
//print("Card Function");
print(s); // print the string itself as an example
// Call the name soup dumping function (LATER, send
// the ep and let the function do the output!
endpoint:Output(endpoint:DumpNameSoup(), nil);
endpoint:Output(unicodeCR, nil);
endpoint:Output(unicodeLF, nil);
endpoint:FlushOutput();
end;
if(StrPos(s, "NAME!", 0)) then // name function
begin
print("Name function");
// Call the Name function (just a wrapper around
// userConfiguration.name)
endpoint:Output(endpoint:DumpName(), nil);
endpoint:Output(unicodeCR, nil);
endpoint:Output(unicodeLF, nil);
endpoint:FlushOutput();
end;
if(StrPos(s, "SIZE!", 0)) then // size function
begin
print("Size function");
endpoint:Output(endpoint:DumpSize(), nil);
endpoint:Output(unicodeCR, nil);
endpoint:Output(unicodeLF, nil);
endpoint:FlushOutput();
end;
if(StrPos(s, "SEND!", 0)) then
begin
print("String sending function, switch to receive string
state");
endpoint:SetInputSpec(endpoint.waitForSTRING);
end;
if(StrPos(s, "BYE!", 0)) then // bye function
begin
print("Bye function");
end;
end,
discardAfter: 200,
},
// our special string handling state
waitForSTRING:
{
InputForm: 'string,
endCharacter: $>,
InputScript: func(endpoint, s)
begin
// blast the string up to someone up there
endpoint:SendOverInfo(s);
endpoint:Output("OK!", nil); // send response
endpoint:FlushOutput();
// back to the main dispatch loop
endpoint:SetInputSpec(endpoint.waitForFUNCTION);
end,
discardAfter: 200,
}
},
DumpNameSoup:
func()
begin
local theSoup, c, val, anErr;
SetValue(result, 'text, "Sending over Card Soup Info");
// get soup
theSoup := GetStores()[0]:GetSoup("Names");
// create a cursor
local c := Query(theSoup, {type: 'index});
// while valid entries, dump the name information out
repeat
begin
val := c:Entry();
if(val.name.class = 'person) then
begin
DumpResult := DumpResult & val.name.last & $,
& val.name.first & $, & val.city & $\n & $\u000A;
end;
end
until (c:Next() = nil);
// return the big string
return DumpResult;
end,
DumpName:
func()
begin
SetValue(result, 'text, "Sending over Name information");
return userConfiguration.name;
end,
DumpSize:
// return size of internal storage
func()
begin
local theStore := GetStores()[0];
local used := theStore:UsedSize();
local space := (theStore:TotalSize() - used) div 1024;
local Kused := used div 1024;
return "K used =" && kUsed && ", K free =" && space;
end,
SendOverInfo:
func(string)
begin
// have our own ref to it (string is changing beneath!)
myString := Clone(string);
// for the time being just draw it in the result view
SetValue(result, 'text, myString);
end,
});
This isnt code to type in, just the params you need to set up the view in the NTK browser. Make sure to click "Allow Access From baseView" for this view named.
//7
result :=
{_proto: protostatictext,
text: "Status Information",
viewBounds: {left: 10, top: 146, right: 226, bottom: 186},
viewFormat: 336,
});