Fix raw type warnings

[marschall]

No description provided.

[oschwald]

I had one specific comment. I also wonder if some of these would be clearer as var.

[oschwald]

Why was this and the other ones that had specific types changed?

[marschall]

Short answer: to avoid compiler warnings.

Long answer:

• DatabaseRecord is <?> here because Networks is <?>. Networks#next() returns DatabaseRecord<T>, so iteration has to use the same type parameter. Otherwise we get either a raw type or unchecked cast warning.
• Networks<?> is the only way to avoid raw type or unchecked cast warnings. The issue here is that Map.class has type Class<Map>. Therefore Reader#networksWithin return Networks<Map> which gives a raw type warning, Networks<?> avoids the warning. I'm not aware of a way to get a Class<Map<?, ?>> or Class<Map<String, String>> without a unchecked cast or raw type warning.

[marschall]

I had one specific comment. I also wonder if some of these would be clearer as var.

I had a look and added a second commit to use var. I'm on the fence here:

• The biggest issue I have that none of the other code uses var. Mixing var and not var in the same method irks me.
• IMO some of them violate the LVTI: Style Guidelines. This applies mostly to the return values of #networks, #networksWithin and #next. Someone more familiar with the code base may see this differently.
• Where the diamond operator can be used I don't see much value in using var instead.
• MultiThreadedTest benefits the most IMO.

Let me know what you think.

[oschwald]

Thanks for the explanations! In regards to var, the primary reason it isn't used more widely is that this code is old and we only dropped Java 8 support a couple of years ago. I realize its usage can be a bit contentious among Java programmers.