how many negatives can I fit in a single sentence

?

8

😄

:feelsbadman:

If 10 people can give a different definition for the same thing it may be...... complected 😎 CSI Miami theme plays

Yeah that all makes sense. Thing is, that’s exactly what I was trying to do! I was trying to lift the convo from implementation into “these are the useful abstractions.” So, for example, in our convo, I tried to say, “CQRS is useful, actually. As proof, you can see many of the relevant abstractions in Datomic.”

But I can see how it seems like I was trying to do the reverse.

“Oh, CQRS is awesome. Datomic implements CQRS.”

I mean, I literally said “But datomic is basically CQRS embodied (depending on your definition).” So. There’s that.

So maybe the takeaway is to be explicit in trying to lift the conversation to abstraction.

Think you about summed it up exactly as it played out haha

To be fair, all this software craftsmanahip mumbo jumbo with Martin Fowler and co. It threads a very dangerous line. I don't know how to define that line, but it's similar to the whole Agile/Scrum and methodologies line

Like, it's easy to devolve in a place which only exists in the clouds, and no one is ever able to really connect back to reality. Where it starts having its own vocabulary of things defined using other made up vocabulary, etc.

I feel I always have to spend so much time deciphering what things mean concretely, and sometimes when you see that part it's like, why does this even have a name

I'm not doing a good job of explaining. But it's like, when people lose sight of the original goals or even problem.

And you start doing a bunch of things for the sake of it.

I think felt the same talking about “Aggregates” with some CQRS people once.

One person was asking, I dunno, something like, “When do you make an aggregate vs allow a read from the write domain object?”

I mean the entire question reveals a certain level of navel-gazing. But it’s easy to ask the question: What do you get from an “aggregate?” Well you get faster reads if they’re materialized (i.e. caching). Other than that, it doesn’t really matter!

Yup, exactly. It's like, what are we even talking about?!?

And when they show you example code, oh boy

ugh yep

It's all MyCoolBlaService with methods etc. The most pendantic OOP stuff

IMO - this is the status quo conversation in software today. Most folks never stop to ask, “What exactly does this get us vs some other option?”

where exactly means, “Matters to non-software people.”

Not to say software people don’t matter. But we spend so much time on ourselves that everyone else gets totally neglected.

That's true

And honestly, I think we can often be off the mark too. Like we get lost in too much abstraction

We lose sight

I read both DDD books, and spent a crazy long time trying to understand, what are you saying, how would I build a system following your descriptions and ideas.

And it's not until I found some other book that I finally understood some parts of it, not all :p

Curious what book made it click for you.

Like how aggregate roots are just supposed to model transactional invariants.

I’ve felt the same every time I have a convo w/ a DDD zealot. I feel like they’re speaking some weird derivative of a language that I know natively. Like Scottish for an English speaker.

Except if you ever try to pin them down, they’re incapable of using normal words, so you never learn anything at all.

Implementing Domain Driven Design by Vernon Vaughn

Haha ya exactly!

Maybe I’ll buy that book, just so I have a translation guide.

Honestly, even after reading this book, it's like... I get it, but I don't lol

That said, it was the best I've found, goes more concrete

The thing is, the model degenerates in practice and that's the part that is never covered

Curious to hear more.

I get how it’s overly-complex out of the box. But not sure what you mean by “degenerate.”

Like figuring out the bounded context is left as an exercise to the reader. And, it's all just intuition

If you end up with the bad partitioning though, you're stuck

Same with your domain entities, if you mess up your aggregates, the whole thing falls apart

Yeah, honestly the whole read model sounds like utter, excuse my french, horseshit.

So let me get a bit more concrete so I don't do what I'm just describing

Take "writes". What even is a write?

We're talking writes to the DB off course

Say we have a SQL db

Writes to a DB are thus done with insert or update SQL queries

Ok, so now we are inside our application

We want to process data before we write it to the DB

Now we're in DDD land. Because it's how we model the data inside our app for modifications inside our app

So this is OOP world. You have for each table a Class (more or less)

Where columns are fields

Ok, say we want to capitalize the first name of User?

Normally you might do getUser(Id), which gives you the current user data, and then go User.setName(user.getName().capitalize())

And then you'll do something like user.save()

Which will generate the query and update the DB

Ok, so what happened here? Did we do a write or a read or both?

uh, both? But I would say, primarily, write (since that was the stated purpose).

We did both, because an update requires first reading in the latest data, changing it, and putting it back

But so this is the first challenge you face. And took me a while to figure it out

So after two books, I realize DDD says

Your domain entities like User, they are only designed in terms of writes

And it's super confusing, but they assume a write requires a read

So when you fetch data from the DB and create domain entities out of it, it should be only for the purpose of making modifications to them

When you don't intend to change them, don't use them.

Instead, when you don't intend to change them, use a different set of entities, which are no longer called your domain entities. Those don't need to be modeled as carefully, just have entities in whatever shape and form your presentation layer cares for

And like, this right there, that's 80% of DDD.

I just summarized 3/4 of all the books I read

lol

I actually had the sneaking suspicion that most folks were lost in some OOP mumbo jumbo.

Anyways, I'm not sure where I'm going with all this anymore :rolling_on_the_floor_laughing:

I guess, too much fluff, not enough substance is my summary :p

I still like the ideas around DDD

Specifically, I like the idea of spending a lot of time to model your data, with regards to how you're going to modify it over time. I like how it encourages breaking out your application in components which map back to the problem space. And I like how it recommends you build a vocabulary that is as close to what the business people use within your code as well, so that communication is improved

I think a lot of the details in the book beyond that are very OOP specific, and you start to fall in that whole complexity nightmare of design pattern and all that

You can try the #yada channel

This doesn't seem to be a bidi question either :)

#clojurescript is your best bet

Is your script tag before or after you load your clojurescript?

<html lang="en"> <head> <title>My App</title> <meta charset="UTF-8"> <meta name="viewport" content="width=device-width, initial-scale=1"> <link href="css/style.css" rel="stylesheet" type="text/css"> <link rel="shortcut icon" type="image/png" href="favicon1.png"/> </head> <body> <div id="app"></div> <script src="js/compiled/myapp.js" type="text/javascript"></script> <script type="text/javascript">myapp.system.go();</script> </body> </html>

Hmm, I think I misunderstood. I thought this was about bidi in clojurescript

Those routes work for me, are you sure they don't work:

user=> (require '[bidi.bidi :as bidi])
nil
user=> (def routes ["" {"/this-route" {"" :index ;; doesn't work
  #_=>                                 "/something" :index-handler ;; doesn't work
  #_=>                                 "/something-else" :some-response }}])
#'user/routes
user=> (bidi/match-route routes "/this-route")
{:handler :index}
user=> (bidi/match-route routes "/this-route/something")
{:handler :index-handler}
user=> (bidi/match-route routes "/this-route/something-else")
{:handler :some-response}
user=> 

The leading slash didn't work. In fact js/myapp.js doesn't exist. The myapp.js that I shared was in myapp-root/dev-target/js and not in myapp-root/resources/js. How do I make a js/myapp.js exist?