Our Databricks-based CDC processing job was lagging during traffic spikes and wasting money during quiet periods. Here’s how we taught it to scale itself.
What I’ve learned about spark streaming checkpointing, how it’s implemented, and how the implementation details of offsets might lead to surprising behavior.
"Testing is still a third-class citizen" The exact reason meteor is disqualified for me. Even the basic manual is broken and wrong in some ways. Internet searches point to frameworks which are broken or obselete and to the meteor testing site (If I recall correctly even docs point to it) which has a notification for a long time about being "rewritten" and no actual content (and would be a paid book .. :( )
If testing is a third class citizen then No real professional software development can take place with it. A real deal breaker
Used this technique myself and I can recommend it, Also Best practices are using only on specific areas of the web page, since this kind of testing is fragile and change prone.
Might be your only option when testing non DOM parts of a webpage (such as canvas based) from the outside (e.g selenium or casper.js for example)
Depending on the project. getting something done faster could be easier with a modern web framwork like Ror or Django (I use Django). But it really depends on what are you trying to build.. Also you should remember that python/ruby enhanced productivity comes with some learning curve, especially for someone coming from java, which is different (not as syntatic but mainly in idioms, best practices etc).
You may want to consider Play framework. another modern web framework that uses scala - which is actually an enhanced and more productive java (and if I recall correctly can compile plain old java as well.. )