The Whiteboard

12:01 AM

a change like that should only happens after discussion with a large chunk of the team, so for us everyone just knows that these files are going to change soon and they'll have to rebase if they touch them

user55340

Its a readers and writers problem. Not an issue if the writers aren't traipsing through the entire package. Its another thing if someone's going to go through with a steam roller.

Ixrec

but then we are at the app dev level, we've never needed to make a major change to our external interface where it has to be done to everything at once for consistency's sake

user55340

I've been involved in several steamroller events. And I've been very glad of file locks.

Ixrec

I usually push to break the steamroller into several more manageable rolling pins

user20683

Anyone feel like savaging my resume?

user55340

12:03 AM

There are times when you can do it in small chunks. There are other times when its "you know what, everything is changing - check in all your code and go work on another project for a week"

Ixrec

correction: for us everyone just knows that these files are going to change soon and any work they do now might be wasted/have to be redone, so there's a pretty strong incentive to obey the non-existent lock

user55340

@Ixrec the best part is when someone checks in a 20% change to one of your files... and then you do a 'vcs update' and find a conflict... oh... great fun.

Ixrec

I think if we tried doing that with actual locks we'd end up with situations where our initial guess of which files will be part of the change is wrong and everyone has to argue for relocking or unlocking files instead of just doing their work

wait, you use vcs?

user55340

change that to 'svn' or 'git' or 'hg' or whatever your flavor.

Ixrec

k

Telastyn

12:16 AM

feature branches ftw.

also, self-referential lambdas are kinda awkward in C#

user55340

I love feature branches. Though that still has the problem of people doing changes and merges to main/trunk/master that you can't really lock down.

Ixrec

isn't part of the idea that it prevents things going into master before they're ready?

user55340

And the 'rewrite this module' type change doesn't work well with merges because nothing is the same anymore.

Telastyn

but it does a good job at communicating that everything is different, and everything is different at one point.

and really, shouldn't the team lead be organizing the work so that the big scary change doesn't conflict with in-flight work?

user55340

Feature branches do prevent the changes to master before they are ready. But you've also got the 'branch bug1234' and then 'merge bug1234 master' and well, it changed some of the code you really wanted to keep not changed.

Telastyn

12:19 AM

or PM if you trust your PM to not suck?

Ixrec

what is this code you wanted not changed? the buggy code?

user55340

The buggy code doesn't exist anymore because it was rewritten.

user55340

One of the more successful (in terms of vcs policy) projects I was on was a 'commit to trunk' approach - everyone committed to trunk all the time. No feature branches. All commits keep everything running. With a lock on about 30 files for a rewrite of that code (that expanded as things where 'oh, crap - that too?' type repercussions).

user55340

It worked. Its one approach to version control... and sometimes I long for it.

Ixrec

I generally do that on simple stuff, like my translation projects where the very idea of a branch would be absurd

user55340

1:04 AM

This question shows part of the fundamental flawed nature of book suggestions quite succinctly:

user55340

14

Q: Is "Analysis Patterns: Reusable Object Models" Still A Good Book To Use For Its Subject Matter Today?

Is "Analysis Patterns: Reusable Object Models" by Martin Fowler (published 1996) still a good book to use for the content it covers? Are there other good books that cover the same content? Are there more up-to-date books? Update: Are there any more answers on this question, as well as any answe...

Ixrec

you should put a screenshot of that in a meta post somewhere

Duga

Possibly off topic. Consider asking at programmers.stackexchange.com — legomaker 44 secs ago

user55340

@Ixrec meta.programmers.stackexchange.com/posts/6487/revisions

user55340

1:19 AM

@RobertHarvey on auto wiring with spring:

user55340

31

A: How does autowiring work in Spring?

Depends on whether you went the annotations route or the bean XML definition route. Say you had the beans defined in your applicationContext.xml: <beans ...> <bean id="userService" class="com.foo.UserServiceImpl"/> <bean id="fooController" class="com.foo.FooController"/> </beans> T...

user55340

262

A: How does autowiring work in Spring?

First, and most important - all Spring beans are managed - they "live" inside a container, called "application context". Second, each application has an entry point to that context. Web applications have a Servlet, JSF uses a el-resolver, etc. Also, there is a place where the application context...

user55340

The way I've been using it is to have my various services all coded with interfaces. Some get left by the wayside as development moves in other directions (I typically start out just doing a 'stub' one that gets moved to the test package eventually).

user55340

And then in the xml config for the application, I say which service implementation gets wired to which interfaces.

user55340

And then suddenly, everything works. Its actually kind of spooky.

Ixrec

1:25 AM

a lot of these Java/.NET technologies you guys talk about sound kind of spooky

user55340

The nice part is that each test can have a different configuration. So I can use that to wire up different services behind it for testing specific code paths.

user55340

class Foo {
    @autowired
    ServiceLayer service;

    public List<Data> doSomething() {
        Map<String, Data> dataums = service.getData();
        return new ArrayList(datums.valueSet());
    }
}

user55340

You get classes that look like that. Few more methods, but its like that. ServiceLayer is an interface.

user55340

And then, there's a class ServiceLayerImpl with the method getData()

Telastyn

bleh.

user55340

1:29 AM

But, it turns out that the ServiceLayerImpl is combing data from multiple data layers. It does a database call, and a web call and combines it... each its own data layer.

user55340

So you have

user55340

class ServiceLayerImpl implements ServiceLayer {
    @autowired
    WebData webdata;

    @autowired
    SomeDatabase db;

    ....
}

user55340

And each of those are interfaces with implementations.

Telastyn

and you hope that all of your N deep dependencies are in your bucket/searchpath otherwise runtime errors.

user20683

anyone want to savage my resume?

user55340

1:31 AM

Yep... though with Spring its a class loader error so you find out real fast.

Telastyn

ah

that's marginally better.

user55340

It does the package scan and such right at the start. Instantiates all the objects and then keeps them in its cache/pool.

user55340

The neat part is that I can have different configurations for different environments... or use the application.xml to inject the proper strings / endpoints.

user55340

I'll admit to always being amazed when I flip that switch and everything wires up correctly. But once its all there, its really easy to work with.

user55340

The current architecture (not full spring koolaid) has a DataLayer.getImpl() type call. Yea, it keeps the layers separate, but doesn't allow for test configurations.

Telastyn

1:36 AM

enh, I wrote something like that for job^^, but only because the environment was insane in the amount of different configurations it needed. That "neat" is an architecture smell to me. The concrete types used in the application should be fixed at compile time (even if they're different for unit/integration tests).

user55340

I was given the opportunity to do a full web app (short time frame, now working on the admin part - which is done, just doing all the testing showing what it can be) and did the full Spring thing.

user55340

The team lead and a new hire are currently fighting with trying to show testing with another legacy code... they're working on launching Jetty (they're not using spring mvc completely/properly) with a database connection to a 'static' database (its not), and then doing String.compare and String.matches tests against the json you get back.

user55340

My code looks like this instead:

user55340

public class TodoControllerTest {

    private MockMvc mockMvc;

    @Autowired
    private TodoService todoServiceMock;

    //Add WebApplicationContext field here.

    //The setUp() method is omitted.

    @Test
    public void findAll_TodosFound_ShouldReturnFoundTodoEntries() throws Exception {
        Todo first = new TodoBuilder()
                .id(1L)
                .description("Lorem ipsum")
                .title("Foo")
                .build();
        Todo second = new TodoBuilder()

user55340

and everything works.

user55340

1:40 AM

No jetty needed - because its using Spring-test. No database needed, because that service is a stubbed service. No String pattern matches because I'm using the ham crest marchers properly because its within a nice junit test.

user55340

Its actually a unit test.

Telastyn

bleh. Builders. But yeah, sounds like your team lead and new hire are not great. Stubbing in a mock DB should be that trivial, spring or no spring.

user55340

That was just some code that I pulled that looked like mine. Ignore the builders.

Telastyn

yeah, not particularly important to the topic at hand.

user55340

Look at the jsonPath validation as part of the call within the mockMvc preform call.

user55340

1:43 AM

The team lead's main problem is he hasn't accepted to work with the framework rather than doing his own thing.

user41796

@WorldEngineer Do I need to be diplomatic about things, or can I just be "engineering blunt?"

Ixrec

I would be very surprised if he wanted the former

Telastyn

nod

to be fair, mocking frameworks do have a bit of a learning cliff.

user55340

Yep. I fought/played with it for two days before I got everything working the way I wanted / it should.

user55340

Incidentally, I found two bugs that I had missed in "click web page testing"

Ixrec

1:46 AM

I think that was too fast for anyone to read it

user15026

@GlenH7 That makes for an interesting experience...

user41796

@Ixrec That was the intent. :-P

user41796

@Ixrec And I figure it's polite to ask. My full bore engineering bluntness has offended some fairly strong individuals. I'd rather not offend World in this particular case as the circumstances certainly don't warrant it.

user41796

But if I don't have to expend the effort of being diplomatic about things, it would certainly speed things up.

user41796

@AshleyNunn It was ... most unusual.

user55340

1:48 AM

Remember, its almost almost pi day. Need to celebrate tomorrow.

user15026

@GlenH7 That's a good way to describe that, I think. Not going to go with that particular one, then?

user41796

The legibility between two hand written coding exercises spaced approximately 15 minutes apart was quite amazing

user20683

@GlenH7 I assume I'm terrible so that's not a great deal that you can burn down

user20683

:P

user41796

@WorldEngineer :-D

user15026

1:48 AM

@MichaelT That explains my friend going "it's almost the one day a year I buy quiche" on Facebook

user41796

No, not terrible. But I'm going to encourage investing in different aspects of said resume.

user55340

22/7 = 3.14285714286

Ixrec

largely by accident, it seems I will be celebrating pi day by going to see Ant-Man

user20683

@GlenH7 mostly trying to figure out more solid way of portraying technical skills

user55340

@Ixrec almost pi day. Pi day is in March.

user15026

1:49 AM

@MichaelT Yeah, part of my brain knew that, but I forget to pay attention. Makes my plans to have pizza for lunch tomorrow all that more apt

Ixrec

@MichaelT rational approximation of pi day?

user15026

@GlenH7 laughs Not surprising, because really, what does one say to that?

user41796

@WorldEngineer Yep, that's what most of my comments will be centered around. I'll work on it tonight / tomorrow. Reply by EOD tomorrow.

user20683

@GlenH7 kk

user41796

@AshleyNunn In the post-mortem, all of us were wondering how it cleared all of the filters we have in place.

user20683

1:51 AM

@GlenH7 I should have a better rig soon, so then I can invest in a windows stack if need be.

user55340

@SimonAndréForsberg You know you want to start up a new Area51 site: ResumeReview.SE... right?

user15026

@GlenH7 That was going to be my question, but hey, some people are good at masking things?

user55340

Btw, I'm enjoying this game: store.steampowered.com/app/317820

Ixrec

@GlenH7 link to the original story?

user15026

@MichaelT I really want to try that one out

user15026

1:53 AM

but I've got so little time for gaming lately, it seems

user55340

@AshleyNunn The nice part about it, its pseudo-casual.

user41796

@AshleyNunn Exactly. Do I play PvZ Garden Warfare tonight or do I catch some extra sleep? Difficult questions.

user55340

You can play a single mission rather quickly.

Ixrec

I basically don't play video games any more except when I have time off

user15026

@GlenH7 The eternal problem :P Sleep, or gaming.....

Ixrec

1:54 AM

one of my explicit goals this week is to play Silent Hill 1 and 3, since I'll never get around to it otherwise

user55340

And the rhymes / puns are ~~awful~~ great.

user15026

@MichaelT That does make it extra appeal to me

user55340

Ok... starting a new 'go exploring' now.

user55340

first encounter done... loot!

Transcript for