Angular framework's grand ambition: Not breaking anything

Angular, the popular web application framework, reached version 4.0.0 last week, having skipped version 3 entirely. Known for its role in the MEAN stack – MongoDB, ExpressJS, AngularJS, and Node.js – Angular lost the "JS" suffix and some of its momentum when version 2 appeared last year, largely because it was incompatible …

[Ilsa Loving]

A platform built on sand

I'm amazed that Google thinks Angular has any right to be considered a platform. A platform needs to be stable. Angular isn't.

You can't just made fundamental non-backward compatible changes to your platform and expect everyone to happily switch over. One of the reasons why Microsoft Windows is so dominant today (aside from the whole convicted monopolist thing), is that Microsoft cared a very great deal about backward compatibility. Heck, if you're using 32-bit Windows 7, you can still run applications that had originally been written for Windows 3.1. That is over two *decades* worth of time where compatibility was maintained.

By comparison, Angular can barely maintain compatibility across 2 years, let alone twenty. That is not the makings of a long-term viable platform. Google says they're trying to "not break anything", but time is the one and only indicator of whether they will hold true, and so far Google's reputation doesn't leave much room for optimism.

[Adam 1]

Re: A platform built on sand

And the counter example of Windows Mobile where they forgot about​backwards compatibility between generations. Developers became disinterested in doing yet another port to a platform with nearly zero user base, then the available apps became a factor with the consumers in a death spiral.

[Anonymous Coward]

Re: A platform built on sand

That's the inherent difference from a business model where you sell operating systems and applications, to another where you give away some code for free to sustain your data gathering and advertisement business (and MS is moving to the latter too, which looks so successful).

You read it: the problem with client side code is not that it opens more security vulnerabilities - it's it can't be "indexed" (aka user data gathered) as much efficiently as if you run it on a server (preferably in a Google cloud, I guess).

I'm always very careful about everything that is a side effect of another, far bigger business. It's clear it will be always driven by the bigger business needs, not its users, especially when they don't pay, so they have no right to complain, after all.

[sabroni]

Re: A platform built on sand

>> One of the major goals with the version 4 release, said Fluin, was to avoid the breaking changes that occurred between version 1 and version 2. <<

So the Angular devs agree with you. Did you read the article?

[Ilsa Loving]

Re: A platform built on sand

>>So the Angular devs agree with you. Did you read the article?

Yes I did. The difference between us is that I am not taking that statement at face value.

Microsoft's "goal" is to make the best Windows ever. We all know how that turned out. Google has a very solid reputation for creating and dropping entire major services like hipster swaps fedoras. Heck, they've gone through what, 4 or 5 completely different messaging systems alone?

The onus is on Google to demonstrate that they can curb their organizational ADD long enough to actually support Angular long term. If they can manage to release multiple versions (not just one) without breaking fundamental compatibility, then maybe Angular will be worth a fresh look.

In the mean time, anyone who actually gives a tinker's damn about future maintainability of the code they write will be better off with tools that have better track records.

[Steve the Cynic]

Re: A platform built on sand

"That is over two *decades* worth of time where compatibility was maintained."

Compare that to a binary compiled in 1967 on a System/360. It can be run *directly* (not in emulation) on a zSystem today. Um. FIVE decades.

[chrigil]

Re: A platform built on sand

To be fair, and I'm not suggesting that this will stand the test of time but the reason for the major breaking changes between 1 & 3 is because v1.x wasn't fit for purpose. It grew out of a pet project (as do so many things at Google) and didn't scale for the new web.

Hopefully v2+ has taken a bug enough step to not require further major changes.

[wikkity]

WebComponents

Angluar is good for want it is good for but goes beyond. Want we need is HTML5 compliant web components supported fully across browsers then we won't platforms.

[Anonymous Coward]

We're having just this debate internally

It's nice having a framework to help with front end development but right now just every project seems to get mostly re-written every couple of years. It means that either you continually rebuilt and re-write work thats already done to keep it upto date or it gets left and then after a couple of years becomes hopeless out of date or worse insecure.

At the end of the day, I've got enough to be getting on with than distracting myself with trying to develop on a constantly moving target.

Besides I thought Google had Polymer as it's pet project?

Anony because I don't want to get involved in the religious front end wars at work

[monty75]

Re: We're having just this debate internally

It's OK. It's not like Google have a track record of creating things only to kill them off a few years later.

[monty75]

"Fluin explained that around 2009 developers began rendering more and more application code on the client-side"

That's a very precise number of developers.

[wikkity]

That's a very precise number of developers.

But not accurate, it's actually 2008, I think someone miscounted