Journal Chacham's Journal: Verbiage: Agile should be prevented, (Java) coders are bad designers. 10
We are using Yammer a the office. Someone started a thread to promote Agile at the office. When my reply got this long though, i figured i should just post a JE already. I edited my replies slightly.
I responded with: "Ugh, i despise Agile . It's an efficient way to deliver trash to the customer in five easy shipments. And forget the data model, it'll be worse than one designed by a java coder. If anything, we should have a group to make sure Agile stays out of our organization."
Someone challenged my generalization, and asked why coders can't design. I responded:
Why should an experienced developer in any technology not be capable of putting together a good data model? Developers and designers approach things differently. Developers architect, designers design.
Developers work on delivering something that meets the requirements. Designers design a model that supports the requirements. Meeting requirements is a perfect fit, but does not scale. Supporting requirements with a model scales.
Developers tend to work from the inside out. That is, get all the points then draw a big circle around it and call it the "model". Designers design from the outside in, draw a circle around an idea, and then check if it supports the requirements.
In Jungian terms (usually), developers are dominant Ts, Designers are dominant Ns. An NT can do both, but one will always be better, because of preference and training.
Designers tend not to code well. What they do works, but lacks runtime efficiency. Coders tend not to design well. What they do works, but lacks scalability.
Java coders are usually the worst of the bunch, as some schools churn them out with low quality. That is not to say that Java coders are bad coders. But bad coders, today, tend to learn Java. Leaving much room for generalizations.
INTP (Score:1)
I like the hobby of programming. My career is in science. Am now seriously thinking about sliding over from perl to python for my open source hobby projects.
I learned a lot from perl since perl 4, since there is more than one way to do it. Now I am thinking that the focus of python on the "best" way will be a refreshing relief from all that choice.
I am enjoying the book "Perl to Python Migration", and "Twisted".
two points (Score:1)
1) Don't take an observation about Java developers and extrapolate it beyond that. Java is like government (and hence prolly why Java is used so much in govt.) -- it attracts the kind of people for whom the priorities are something cushy and not having to really care about anything. I.e. people go into Java because it promised to be relatively easy and to wipe your butt for you when you mess yourself, and these kind of people naturally aren't interested in learning about all kinds of technical things in any
Re: (Score:1)
1) That's my point!
2) Coder to not design data, they design the application. Two completely different things.
Re: (Score:1)
2) Coder to not design data
What do you think the programming topic of data structures is all about then? Or file layouts? Or marshalling domain "things" across sockets? What do you think half of object-oriented analysis and design is? What do you think coders are doing when they create an XML schema, or a WSDL document? There's a heck of a lot more to designing data than just what's under the database's purview.
Re: (Score:1)
I think we're getting into an issue of nomenclature.
Data can be stored either based on what it is, or based on how it is used. A data model stores it based on what it is. An application stores it on how it is used.
An expert at one, is unlikely to be an expert in the other. That is what i want to say.
Re: (Score:1)
But in the real world isn't it exactly both? As an example from object-oriented modeling, in OO Analysis you examine the problem domain and determine a set of candidate objects from it, but when you put on your OO Design hat you typically change some things around. It seems to me that not factoring in how data is used, implies data that's never used, so then why store it in the first place.
Re: (Score:1)
But in the real world isn't it exactly both?
Nope, far from it.
Look at data models designed by coders, and look at data models designed by designers. The coders model matches its use, usually process oriented, and is patched as needed. There is usually some redundancy or a lot of confusion in using one that's been around (and been patched) for a while. The designers model matches an idea, usually object oriented, and it's expanded (as in making the circle wider) if needed. Usually little expansion is needed,
Re: (Score:1)
The models i have designed, stay longer.
And the work I've done has to be revisited far less than what other coders produce. So maybe it's not that coders can only design code well and can't design data well, but that in general they can't do *anything* well. Furthermore:
On the last significant (multi-year) project I was on, the "data modeler" who was originally on the project and prior to my joining the team did things like model a polygon of coordinates by placing the first coordinate of each one in the po
Re: (Score:1)
My experience is stupidity is an equal opportunity striker
Heh. Unfortunate, but true.
and that most people suck, period, no matter what their job title is. And that a few, such as possibly you and I, suck much less, because we're actually trying, and give a hoot.
True as well. But there are still natural skill sets.
placing the first coordinate of each one in the polygons table
Storing conveniently is usually a bad idea. Unless it is a default. Like a default translation, or a default price. The first of a list