Google Trends: Javascript Frameworks

It's nice to hear that someone else finds Ext's choices hard to understand at times, I'm the only guy around here who needs to work with it so it can be hard to know when it is me and when it is something fishy.

Don't get me started about the almost completely undocumented "Sencha Architect" which overwrites any changes you make to code in another editor when you re open a project....

I spent 2 years writing against the latest stable ExtJS. At about 3 months, I stopped using their docs for anything but cursory information: what methods does this class have of its own, what does it inherit, what is its class hierarchy, etc. That information is dynamically generated and typically spot on. The rest is generated by doc comments, and it's very hit or miss.

Instead, if I needed to know ANY details of a function, I clicked on the '.js' link on the right side of the page in the class hierarchy to load the .js file in the browser, then a quick cmd+f would show me what I needed to know, because I can parse consistently formatted JS faster than I can grok their vertically challenged documentation. JSDuck pulls inherited docs if function-level docs are omitted in a subclass, too, and sometimes they're just plain wrong because of that.

After about 6 months, I had committed to memory about 60% of Ext (all but the graphing bits, mainly, which I never used). I believe a good library is one that you don't need to worry about, ever. So naturally, being required to delve into Ext's source so often to me is unacceptable of something that is a paid product.

I would've been fine to fix so much in Ext, but it isn't developed in public via something like github, even though it's dual licensed, which meant the only way to do that was to post on their forums. No thanks. 4.x came a long way in rendering performance, but there were still critical paths that were untouched, including layout logic that determined when to perform layouts. They aren't required almost ever, but are used liberally, so while making the process faster is a benefit, the fundamental problem remains. A 200ms (for us, YMMV) layout problem was fixed (down to < 1ms) by a +5-0 change that added heuristics to determine when a layout was actually needed. This is a trivial change, and could be thrown in a PR at no cost. I'm not going to copy my code on the forums and hope someone at Sencha will manually merge that into whatever bleeding-edge version they have.

Note, this is all anecdotal- My adventures with direct proxies and nested layouts had me doing a lot of trial and error coding, this may not be a representative experience.

Basically I built a dal access layer based on promises for delegating to from my viewmodels. I also built a view transition manager into my app shell also, part of my core SPA framework.

I also added a bespoke option piece for hydrating routes. I have this api for my SPA called /public/routing and user/{id}/routing which returns the json config for API's for either anonymous or authenticated users. it is key on the Sammy Hash intercepts and provides me with the glue, i.e. what external view template should I transition too, what view Model should I load for this, also what API and Verbs are supported for the users role! its great, because my site admin can added routes on the fly, and its totally modular from a dev point of view, so if a new route can not be found, I just load the Empty View Template Placeholder (in development environment of course ).

But I agree, extjs needs a considerable amount of investment in time and mindshare, because of the way it is built. I actually like it very much. Hey I built my product based on it.

But lately I have seen extjs as old, and in need of love. AngularJs for example is easily lovable and easy to learn.

But yes, I use it for corporate software, since my company pays the license. Also it's strong in the table and charting parts, which are the butter and bread of many corporate devs.

So the keyword still applies. Search for sencha might be mixed with other frameworks they have.

Now these trends focus on a world wide searches. If you look closely at the detail, the number of search for extJs in China, pretty much take the 90% of the searches. And in the U.S. extJs has practically no prescence.

If you look at the detail for AngularJS, most of the searches are in California, New York and India.

So it is really not the best trend analysis, but it does put the power of China into perspective.