Rob Tweed · Sep 28

glsdb: JavaScript Objects that are actually IRIS Objects

I'd like to announce the release of something really rather interesting - revolutionary in fact.  That may sound like hyperbole, but I don't think you'll have seen anything quite like this, or even thought it possible!

We've pushed out a new JavaScript/Node.js module named glsdb which you can read all about here in detail:

However, for the purposes of this announcement here, I just want to focus on one part of glsdb: its APIs that abstract IRIS (or Cache) Classes as equivalent JavaScript Objects.

Let's take a quick look at what I mean by that, because I mean JavaScript Objects that are actually IRIS Objects residing in the database! 

Suppose I've installed the SAMPLES data from the Intersystems repository, eg as described here:

So, to access this SAMPLES data using glsdb, I first have to open a connection to IRIS (glsdb uses our mg-dbx interface for making that connection), for example:

const {glsDB} = await import('glsdb');
let glsdb = new glsDB('iris');
let options = {
  connection: 'network',
  host: '',
  tcp_port: 7042,
  username: "_SYSTEM",
  password: "xxxxxx",
  namespace: "SAMPLES"

Now that I've done that, I can do the following to access an Employee record in the SAMPLES database:

let Employee = glsdb.irisClass('Sample.Employee');
let employee = Employee(101);

So far, so unsurprising, perhaps.

But here's where the mind-boggling fun begins.  We've created a JavaScript object named employee which appears to represent the Employee instance with an OID of 101, but it's actually a rather special JavaScript Object: it's a JavaScript Proxy Object that is directly mapped to the physical IRIS Class instance in the database.  I have direct access to all the IRIS Class's properties and methods, and when I manipulate the Proxy Object, I'm actually manipulating the IRIS Class instance in the database!  So I can now do things like this:


let salary = employee.Salary;
console.log('employee.Salary = ' + salary);
let company = employee.Company;
let name = company.Name;
console.log('Company: ' + name);

 The important thing to realise here is that my employee object isn't an in-memory JavaScript copy of the IRIS Class instance - it is the actual IRIS Class instance, in the database!

And as you can see above, I can even chain through the relationship between the Employee and the Company as you would in ObjectScript.  In fact I can even do this in one go:


and chain directly through the properties, directly in the physical IRIS Class in the database!

I can save changes to the object too - for that I use a special method named _save(), eg:  employee._save()

To create a new instance of an employee, I'd just not specify an OID, eg:

let newEmployee = Employee();

glsdb provides a shortcut way of defining and saving a new record all in one go, using the _set() method, eg:

let ok = newEmployee._set({
    Title: 'Manager'
    Salary: 60000

You can even inspect the methods and properties available for the Class:


So, that's a quick look at glsdb: I hope it gives you a taste of how crazy it is!

You're probably wondering how is it even possible to do this kind of thing. 

Well, in summary, it's all down to a relatively new feature of JavaScript: Proxy Objects.   Proxy Objects are special objects that act, as their name implies, as a proxy to another real object.  But their real power is in the fact that you can set traps for access or change to the Proxy Object's properties, and (and this is the key bit)  you can apply custom logic to those traps.  In the case of glsdb, that custom logic uses the mg-dbx APIs behind the scenes to perform the equivalent access or change that's happening to the Proxy Object to a corresponding IRIS class.

Anyway, that's a quick glimpse at the crazy world made possible with glsdb!  What I've put out is an early release so I'm sure there are aspects of the IRIS proxying that will need tweaking.  if you want to try it out and see what it can do, please let me know if you find anything that doesn't behave properly. 

Oh and by the way, check out the other APIs that glsdb provides - they're equally interesting and mind-boggling!

And finally, glsdb is Open Source software, so please play around with the code and post PRs on the Github repository if you'd like to help out with its onward development.

1 162
Discussion (7)3
Log in or sign up to continue

How intriguing Rob.

May I suggest you edit the post and change the language at the top of the CodeSnippet elements so they're set to JavaScript instead of ObjectScript? Or if you prefer, ask me and I'll do it using my DC Moderator powers.

Thanks for pointing that out, John.  I've changed them :-)

Impressive, thanks so much sharing that with us.

Few questions come to my mind :

  • There is no example for IRIS in you github do you plan to do it ?
  • Are you using behind the scene the NativeAPI from IRIS ?
    • If answer is true, are you using the Calling ObjectScript Methods and Functions ?
    • Else, why not ?
  • There is no example for IRIS in you github do you plan to do it ?

Yes there is if you look (I do need to add an index to the Readme):

If you mean a worked example?  No - not yet but I'll add a simple one in due course.

  • Are you using behind the scene the NativeAPI from IRIS  ?

No, as the documentation says, we use our own mg-dbx interface and its APIs:

This is designed to get the very highest performance through the Node.js API (which otherwise has serious bottlenecks).  It's also compatible with most earlier versions of IRIS and Cache and with Node.js v14 - 18, as well as aligned with our other similar interfaces for use with other technologies.

mg-dbx provides all the APIs needed to access IRIS classes:

One point I should mention - this kind of direct abstraction of IRIS objects onto another language's in-memory data structures (JSON/JavaScript Objects in this case) would normally have been thought to require the language to be embedded into the IRIS (or other database) environment (and therefore requiring implementation by the vendor).  glsdb demonstrates that it can actually be done without such embedding - it's all done using standard Node.js language features, and can even be achieved over a networked connection between Node.js and IRIS.  And with mg-dbx providing the interface, performance isn't sacrificed either.

... which makes me wonder about a Python equivalent of glsdb.  I notice that Python includes Proxy Objects which I assume work similarly to JavaScript's Proxy.  Our mg_python interface provides all the same IRIS APIs as mg-dbx.  I'm not a Python developer, so wouldn't be the right person to implement such a beast, but all the logic that's needed is in my glsdb Github repo for an appropriately-skilled Python developer to translate from JavaScript.  All the features of embedded Python but using standard Python :-)

Rob, this is a great addition to the JavaScript (Node.js) tech stack for IRIS because glsdb allows you to work with classes and globals as if they were simple JS arrays or objects. A JS developer has now persisted arrays/objects at his disposal in a very simple & familiar way.

This is a big step forward for creating Node.js back-end applications with IRIS as the underlying database when you combine this with any standard JS front-end framework + tools.