Martin Farrell

My issue with GraphQL over REST is that with a REST service I can have an optimised query to a database to retrieve the data as efficiently as possible. But when I move to a graphQL the client can then pick and choose what fields they request and retrieving the data requested may be more efficient running different queries based on those selection.

I ended up using node-postgres since im generally talking to microsevices with smaller schema's and you can push code faster

Great article. A lot of the problem with "Keeping up with technology" is that the underlying ideas dont change that much on a year to year basis, but you are learning new API's or syntax to solve the same problems.

Its like someone rearranging my garage a few times a year, and i spend the next few months finding out where they have put things before starting again

I thought -

management.endpoint.shutdown.enabletrue

Would only enable the shutdown endpoint?

Problem with graphQL is that its not as simple to use as REST - you have to structure a GQL query/mutation to access the data. While REST is just call an endpoint - then use the data. I also think REST is easier to document

I can see why a high volume service like FB would use GQL - but for smaller sites I think REST is simpler

One thing I've found with microservice development is that I'm not mapping entity models as extensively as I once did with say JSF development. With the MS model it makes more sense to map the key relationships I need for each call, then use spring data repositories for access, rather than traversing a JPA model. Has anyone else found this?

I think we'll see abstract classes used less in favour of interfaces with default methods

This is a really good post - Im inclined to agree with you on MRJAR's being a bad idea. Just seems to pollute a codebase

I might look at this too for an evaluation of options for spring boot

Probably needs profiling - one of the advantages I see running this as a service is I can schedule a daily restart to clear memory leaks if any. It might be an unpopular approach - but it buys time to solve the real problem

i dont think so - you can do that with procrun but its harder to implement

i dont think so - you can do that with procrun but its harder to implement

Im not sure I would run VisualVM in production - pre-prod environments yes, but not production

as a general wrapper or with spring boot?

yes - i normally have a directory per deployment

I always liked the sql taglib on JSP best

Nice article, but Im coming to the conclusion that external tables are a better way to load CSV's into databases than Spring Batch, even in circumstances where we want to process the data once loaded. Its simpler and quicker in my experience

Well if you mean it returns all elements that satisfy the predicate, and in the case of ordered lists this is the elements to the left of the predicate.

What you are asking for is dropWhile - the opposite -

Stream.of(1,2,3,4,5).dropWhile(p -> p < 4).forEach(System.out::print);

Good post - is the reasoning something to do with the lambda compiling to a static method, so making all Exceptions unchecked means the calling methods signatures dont have to be changed?

Thanks Oliver. I based this post on a genuine issue I had and needed to expose the refresh method. The reason for doing it like this was this post http://forum.spring.io/forum/spring-projects/data/104364-entity-refresh-using-spring-data-jpa-module where you advise to expose refresh on a Repository.

Its a 2011 post, so things may have changed, but I felt it was a reasonable approach, but thanks for your feedback

Nice article - documentation marginally better in Java 9 - https://docs.oracle.com/javase/9/docs/api/java/util/Formatter.html

Thanks - Ive fixed the typo

It looks like it - i wonder if there are any nuances

I was looking at this project last week as a spring version of django admin - looks a good project. I couldnt see how it would work at a microservice level as you would need it to be more centralised

I dont understand how its passwordless? You need a password to access your email - so its more a delegated password. 2FA sounds better

My next post is conclusions and design considerations of using Spring Data REST, and one of the downsides is the level of access that exposing a Repository leaves, and whether you are better using a DTO approach.

You can also override default methods in interfaces -

interface Def1 {
public default String getString() {
return "Def1";
}
}

interface Def2 extends Def1 {
public default String getString() {
return "Def2";
}
}

Its also worth a look at procrun - https://www.javabullets.com/procrun-java-programs-as-windows-services/ - the advantage being procrun is used by apache projects

How does this compare to hibernate-envers?

Good summary. I read the discussion on Hackernews and a lot of people were implying that IBM and JBOSS were purely motivated by politics and support of OSGI. But if you read the JBOSS blog the main issue was backward compatability being lost through Jigsaw, and their no vote was to allow time for this issue to be addressed. This was also the main objection of the London Java Community

Good article - I would have used the Spring Data framework for persistence - your implementation just leaves it to the implementer to decide. If that was the case I would have considered moving the data access to your PersonService?

Yes that was the specific problem they wanted to solve - how to introduce API changes to an API structure that may not be under your control - and they would have struggled to solve it any other way. My concern is that default methods will be prone to misuse - but then again thats how languages evolve

I cant decide if default methods are a great step forward or potential hazard. It depends on the usage as the use case for default methods is backward compatability in API's. I fear it will get misused

Good explanation - but you can understand the confusion -

Java - One per Classloader

Spring - one spring bean per id per spring container

I was using this approach recently with spring security its worth a look on configuring 2 levels of security

Good tutorial. Got a bit confused by the javax.enterprise.event.Event.fire method - and was trying to place it in the FireStation concept in your example, not the Event object

Its not purely OO - but pragmatic OO - like you observe it uses primitives for compiler optimizations

disappointed you used amazon referral links

I dont think its that big an issue. Stackoverflow is the place to go to get quick answers to smaller problems - an example being best practice in a new version of a library or language, or to clarify syntax. But the moment the problem gets more involved then you are back hunting for blogs or reading the manual to understand how to implement the code. That said the best posts on stackoverflow are in the form of blog posts.

Thanks for your feedback - The point i was trying to make is that to use this approach you must associate the annotation to a JPA object

I can work with both - but prefer spring simply because i can hit the ground running with it. You have projects like -

> spring boot for a quick start

> spring data allow you to quickly define your core DAO's

> Spring Loaded for hot deploy - you can get this for JEE but you have to pay

I also agree with the criticism of JSF - its ok but i find it too abstracted

I have the same issue with Lombok that you have cited - i like to know whats going on. The annotations seem like a good idea, but I like to see my equals, hashcode's and toStrings. I also think its important for developers to see this code as they apply a number of important rules in Java. With this in mind I just let the IDE create the methods

Id also include androidkickstartr - great to get a base application integrating a number of the above projects (robojuice, sherlock)