The Whiteboard

Jimmy Hoffa

12:04 AM

Mumble this cold is kicking my ass, I think the wife's going to do most the trick or treating

Robert Harvey

They have these things called coats.

Jimmy Hoffa

I mean rhinovirus not temperature, as a kid in Colorado every costume had to involve a winter coat and snow boots

Not the same these days

Jimmy Hoffa

12:58 AM

@MichaelT I gotta say...c'mon, candy canes? Don't you remember trick'r'treating? Candy canes... :P We stock up on all the best (palm-oil free!) chocolate bars that everyone loves and everyone gets to pick 2 or 3 from their bowl. I like to be the house I liked trick'r'treating at

btw I told my wife about your M&Ms->Skittles gag the other day and she about fell down

user55340

1:25 AM

@JimmyHoffa The kids actually seem to appreciate it to a degree. Its not something they normally get in the bag. I mean, how many small bags of M&Ms can you get before you're bored of them.

user55340

I'm also the one that they remember.

gnat

3:26 AM

gnat

6:48 AM

@MichaelT I just made sort of a follow up to your recent question edit debate...

this post addressed only a part of the question asked - a troublesome / resource request / off-topic part of an otherwise reasonable question. It's barely more than a bunch of links, lacking an explanation and context, hardly complying with How to Answer guidelines. In debates like that I am usually on the side of the answerers, up to preferring question closure over editing, only to preserve existing answers. However, quality of this "answer" literally pushed me to opposite side - a question edit that invalidates only an answer like that, must be a good one — gnat 1 min ago

thorsten müller

10:27 AM

Has anybody here some experience with the DCI approach? Is this any good or just another new buzzword? Just reading Clean Ruby by Jim Gay and wonder if this is any good for some medium complex web api project.

Jimmy Hoffa

1:51 PM

@gnat That's hilarious.

user20683

2:34 PM

@JimmyHoffa This fresh grad is still on the lawn :P

Jimmy Hoffa

@WorldEngineer At least you cut it

0

Q: MSM Merge Modules in Visual Studio 2013

Could someone please let me know where I might find resources for creating MSM files? While I am able to create MSI files using InstallShield, it seems that Visual Studio no longer supports Merge Module Projects, judging by the link below and the screenshot of my version of Visual Studio 2013 - h...

c# visual-studio

gnat

@JimmyHoffa full text is on google code if you're interested: code.google.com/p/xee/source/browse/XeePhotoshopLoader.m#102

            // At this point, I'd like to take a moment to speak to you about the Adobe PSD format.
            // PSD is not a good format. PSD is not even a bad format. Calling it such would be an
            // insult to other bad formats, such as PCX or JPEG. No, PSD is an abysmal format. Having
            // worked on this code for several weeks now, my hate for PSD has grown to a raging fire
            // that burns with the fierce passion of a million suns.
            // If there are two different ways of doing something, PSD will do both, in different

a piece of code I maintained once went through review like that. It turned out very educational and helpful

thorsten müller

Anybody any experience with DCI? Just reading the book by Jim Gay and wondering if it's worth to use it for a medium complex web api.

user20683

2:50 PM

Just a heads up, Stack Overflow is less than 1k away from 6 million questions.

user20683

less than 900 now

user55340

@thorstenmüller Haven't heard about... what does this TLA mean?

user20683

@MichaelT data, coding, interaction I think

thorsten müller

Data, context and interaction

kind of expansion of MVC

Trygve Reenskaug came up with it so at least that much credibility to it. People discussing it maybe two years now. Though I didn't follow that too much. But at least part of the Rails community is hyping it quite a bit.

Jimmy Hoffa

3:11 PM

@thorstenmüller Yeah, but how much of the hype from the Rails community is deserving of any amount of trust..

;P

Really though, I'm curious. Care to explain it a bit @thorstenmüller?

If it's been meandering about a community for 2 years it's slightly more than a buzzword. 2 years is longer than any of these *DDs have been around

(other than TDD of course)

thorsten müller

Yes, we do a lot of such stuff only to annoy the PHP guys. "Look how easy this is in Ruby. How many LOC do you think it would take in PHP?". Well, in addition Jim Gay found his worth his time to write a book about it.

The main idea is to split classes and remove functionality that is only used in some specific context into "related" classes. (these are mainly the model classes)

Robert Harvey

Where is Yannis' meta post with the site growth graphs?

thorsten müller

So a class like an account has use cases or roles it fullfills only in a soecific context. And you somehow "attach" the functuinality neede for this role at run time. (Which is so easy in Ruby). That way reducing the overall complexity of a class for the programmer.

Jimmy Hoffa

3:33 PM

@thorstenmüller Not sure I understand your description

sounds like it's written as an anemic domain model, which at run-time has the anemia solved by adding the guts back into the objects using ruby's dynamism

thorsten müller

You can do this in C++ too it seems, but yes, Ruby's monkey patching comes in nice here. The basic classes are reduced to keeping "state" and then are filled with functionality by the roles.

In Rails they are talking about this thin controllerm fat model thing. Then models become a bit too fat for some people taste and this would be one way to solve it.

I'm tasked with moving our web site into a web api for the coming months. So I somewhat face exactly that problem and looking for some options. Just not sure if this really helps or just explodes everything into tiny classes.

Jimmy Hoffa

@thorstenmüller This just sounds like making every method a static extension... nothing really wrong with it, just seems odd. The only reason I know of for taking logic out of domain models is when you want to decouple it so the logic can be used across multiple types rather than just that one type, but in your case you don't get that benefit, though you still get the cost of everytime you change the domain model you have to root around for the methods that use it to update them too...

Not sure I like it.

user55340

@RobertHarvey have you found it yet?

Jimmy Hoffa

I like separating logic and data, but the reason I do this is so that my logic can be used on multiple types rather than only one, but I'm guessing that's not occurring in the case of this approach you're talking about, so the separation gets you nothing really.. I guess it's so you can put in variant sets of logic depending on the 'role' you're in?

@RobertHarvey Just check @YannisRizos' MSO answer history

Robert Harvey

Easier said than done.

user55340

3:47 PM

As seen from the stats in why are the best questions the ones which have been closed? on MSO, P.SE only started gaining in popularity after people started actively moderating it - closing questions and deleting things that should be deleted. — MichaelT 21 hours ago

thorsten müller

Yes, that's exactly my point. At least in its 'pure' form it seems at least to add a lot of extra classes and files. There are cases where you can reuse some roles of course. Especially in case of an API where those roles can be rather abstract functionality.

Robert Harvey

@MichaelT Ah, it's not on the Programmers Meta. That explains it.

Bookmarked.

Jimmy Hoffa

@thorstenmüller Yeah and when you have that abstract enough functionality that they can be generalized across multiple domain models, it makes sense. But that's a tiny minority of cases in OO systems. Other than those scenarios you're just separating logic from data just to separate them even though they're still tightly coupled and still need to be updated in synchrony due to that coupling

user55340

@RobertHarvey And also likely where you noticed it - that comment would have shown up in your inbox.

thorsten müller

They are talking a lot about 'concentrating on the current task'. So it is aimed a lot towards the programmer and only having to think about the things you want an object do right now.

Jimmy Hoffa

3:50 PM

Again I could see Role X puts functionality A,B on domain model K, but Role Y puts functionality B,C on domain model K, but then that also doesn't seem right to me

Robert Harvey

@MichaelT There was a time when I knew everything about the SE network, and what was happening on it. All of the meta conversations, every website rule and programming intricacy. That time has long past.

thorsten müller

It somehow breaks encapsulation, since the role classes have to access a lot of data.

Jimmy Hoffa

@thorstenmüller I'm not concerned with encapsulation too much, encapsulation is just type coupling, I just don't like that they're doing something which is meant to break type coupling apart but in a language which still relies largely on type coupling where the designs are going to be type coupling

It sounds like they're trying to do what Type Classes do

You write your type classes logic implementation somewhere separate from your data definition, but the reason is that type class can be consumed by others whether your actual type is a list or either or maybe

thorsten müller

Interesting point. In some sense the whole "duck typing" thing has some similarity. But in this case I think it's a bit different. Though you would have to know the Rails community and maybe some specifics about web development.

Jimmy Hoffa

In Ruby I'm not suspecting when you create some bit of logic like "Save to database" or whatever it is you do, it's to be abstracted into a role that can be applied to a bunch of different objects, because that's just not how OO approaches stuff

If you can abstract it so that role can be consumed by people not caring whether they are acting on an underlying account model or car model or duck model, then there's value in separating the logic from the model, but so long as the consumer is tightly coupled to the actual model, then you lose that abstraction

thorsten müller

4:03 PM

Yes, but Rails devs monkey patch all the time. That's how the whole Rails thing works after all. This is by no means a CS oriented community.

Jimmy Hoffa

@thorstenmüller Shrug, I don't actually know squat about rails, maybe it does make sense. It seems strange to me to see it useful in an OO approach because type coupling is part and parcel of OO, but then I don't know how strictly OO Rails crup is

thorsten müller

The whole reusing code aspects doesn't really play a role in this discussion since this can already be handled by more common OO patterns.

About the strictness that's difficult to say. On one side the whole Ruby thing is all about conventions following 100% best practices. Up to the level that really every single Ruby programmer indents with two spaces (because Matz says that's good)

On the other side it's all about this amzing flexibility of the language and how you can reduce everything to a few lines of code (because the heavy lifting is hidden behind the scene that is you actual controller)

So how strictly OO can a framework be that allows you to call methods that where never defined and that are created at runtime by splitting the function name into pieces?

Somehow anyway, the method to do the splitting and generate SQL from it must be defined at some point obvioulsy.

user55340

@RobertHarvey Its a much larger site/network than it was. I remember on other sites where I knew everything about it... and years later, only knew a small fraction of what it had grown to since.

user55340

And Mods, I'm sorry about that difficult flag (bounty) that I just did - I know it puts you in an awkward (evil mod) place.

Sparticus

Came in to find everyone talking about Ruby, how refreshing

thorsten müller

4:19 PM

I just try to find a way to align the Bruce Lee "Be like water" quote with Ruby programming.

user55340

@Sparticus I've been holding off on my rants...

thorsten müller

And the find another quote that fits the theme and says something bad about PHP at the same time.

But if I would have any success Yanis would ban me from chat anyway.

user55340

@RobertHarvey (btw I feel kind of bad giving a link to monop from ~1980 C (with code that has #ifdef (pdp11) in it and a custom implementation of malloc)).

user20683

4:54 PM

70ish questions to go on SO to 6 million

Transcript for