What is the present tech landscape?
If we look at present web development technologies, we can see everything is revolving around client side HTML generation. The interesting thing at client side web development is that, there are new tools and frameworks getting released every week. It is very difficult to select one and get something out using that. By the time we complete analysis on one tool or framework, the next one will be out by saying its better than the existing. If we keep analyzing, we never will be able to do something useful.
In short, I selected Angular JS 1.x over other technologies even by avoiding Angular 2. This post talks about my journey towards Angular JS 1.x to do this time bound task over Angular 2.
There is no standard tooling for ng2
We can easily compare the scenario with the US restaurants v/s Indian restaurants.
In India, if we got o middle or lower level restaurants and order Masala Dosa, they will give us a standard Masala Dosa. The standard changes from state to state and district to district. Very less options to customize. But if we are in American restaurant, they will ask so many questions, if we order a sandwich. What type of cheese, what ingredients, type of bread etc...
But in Angular 2, there are many tools and we have to pick up the combination what works for us. What framework for module loading, how and when to compile TypeScript to JS, how to do packaging. New tools are popping up frequently too. Though there is official recommended tool chain, I don't think it is working out for many developers. May be they are not productive or are complex which needs more technical knowledge. Else people will never spend time to invent more and more tools.
Though the ng2 framework is kind of mature the tools are not.This may be controversial statement. There will be many disagreements as some tools works for some and all can find some combinations.
Many tools are in beta
The main tool came along with ng2 is the command line tool called Angular cli. At the time of writing this post, it is still in beta. They recently switched from SystemJS to WebPack.
Some tools doesn't mean for WindowsSuppose we take the risk of using a beta tool for doing development, there are still issues. Some are more tailored to OS other than Windows. Though we can mitigate the issues, it eats our time.
Compiling TypeScript in browser is for demos till browsers understand TypeScript
Now a days most of us agree that the compilation of TypeScript needs to be done along with build process. Though there are options available to compile TypeScript in browser, it will reduce the start time of our application. The other group is telling that, some browsers are expected to support TypeScript directly in future so the compilation can be done by detecting the browser.
This is same, if we select ES6 as well. Some says we should compile down to ES5 always the other side argues that it should be done at Browser by detecting the feature support.
Its all good if we are developing a demo to get a project or samples. But for something to be done in a time based manner, we cannot be trapped in these arguments.
Why not code Angular2 in ECMA 5 JS?
Come on...Though there are many good tutorials available which explains writing Angular 2 application using ES 5, I never want to do. There are many reasons. One big reason is going forward, if we struck in an issue the sample snippets or solutions provided will never be in ES5. We have to find out how to convert the ES6 or TS code into ES5.
But we should definitely try writing the Angular 2 code in ES5 for understanding purpose. Else if something goes wrong in our compiled ES5 code, we may not be able to find out.
Why not build a brand new site and switch one morning
This is a famous question we all might have faced during any migration effort. The developers who never tried rewriting a software under proper timelines, will always advocate this strategy. But others never agree on complete rewrite. When we look at any software which is running for long, we can see so many pain points of previous developers and hacks, anti patterns etc...If we just start rewriting, we will also have to go to the same pain points. So why should we take unwanted risk? Lets always divide and conquer part by part.
In my personal web site scenario, I was the only developer and I am the one who is going to rewrite also. But I had taken so many things from here and there during initial development and unfortunately, I didn't document anywhere even in this blog. Since I use my brain mainly for generating ideas, I don't remember all the pain points now. Hence there is no need to take risk.
Again why should we take same old risks as there will be new risks during the conversion to client side. By this time I already got an issue related to Angular date formatting in Windows Phone 7. Please don't ask apart from me who else using Window Phone.
The continuous delivery system
I currently use free service from AppVeyor to run continuous delivery of my web site. Since its a Visual Studio solution, AppVeyor detect it, compile the same, run tests, package and deploy to the server. To my best understanding ng2 heavily depends on NPN though there is a nuget package for ng2. AppVeyor can be configured to use NPN, but it is again time consuming.
When is the best time for ng2
None of the tools and technologies are going to be there for ever. The technologies are suitable to the current landscape. There are new cool technologies coming out. Below are some
- ReactJS (This is already out. Though it is not a full fledged framework like Angular, it is a choice)
- WebAssembly (we may be developing browser apps using C#. Remember the old Silverlight days)
The driving factor in this conversion is the cost to host ASP.Net site. After that based on the situation, there may be pressing situation to convert to any one of the above technologies. But right now, it is impossible to predict.
These are my own views about selection of AngularJS 1.x for a personal/portfolio site. These are not applicable to all the scenarios.