ISerializable - Latest Comments

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Satellite Phones — Thu, 25 Feb 2010 14:04:56 -0000

I'm a software developer, building applications in FileMaker Pro and I'm trying to decide on the best strategy for tracking my work.

regards
james
______________________________________________

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Baltimore Heating — Thu, 25 Feb 2010 14:02:15 -0000

You seem to have got the niche from the root, Awesome work

respect
raven conway
______________________________________________

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

catering company raleigh — Fri, 22 Jan 2010 08:47:52 -0000

I'm a software developer, building applications in FileMaker Pro and I'm trying to decide on the best strategy for tracking my work.

Have a nice day
jamrson
______________________________________________

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

product testing services — Sat, 22 Aug 2009 13:47:56 -0000

test

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Omar — Sat, 11 Oct 2008 16:01:24 -0000

How difficult/time it will take to adapt unit-testing to developers?
How to verify/assert that developers are indeed writing tests?
How to measure quality of tests written by developers?

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

john — Fri, 27 Jun 2008 15:57:38 -0000

can anyone please write unit tests for the following method

public string RemoveRepeat(string Param)
{
string NewParam = string.Empty;
foreach (char c in Param)
{
if (NewParam.IndexOf(c) == -1 )
{
NewParam += c.ToString();
}
}
return NewParam;

}

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Thomas Jespersen — Sun, 25 May 2008 19:21:45 -0000

6 Should a team be forced to use TDD/Unit Testing
A:There should always be at least one person on a team, how relay believes in unit tests/TDD, and have the position to take a leading role. If the entire team of developers do not believe in TDD/Unit Test should be forced to use.

7 Doing TDD and making you code testable requires a lot more code than "the traditional way" (Interfaces, Mock/Stubs, Unit Tests etc.. Doesn't we risk over engineering?
A: Personally I think the answer is yes to both questions. I know that this is controversial viewpoint in this forum, but it seams to me that no one stops, and asks if its all worth it. When doing TDD one easily ends up writing twice as much code. Code that has to be maintained and cared about.

While the tests makes it easier to do refactoring, it also ads to the amount of code that needs to be changed, if you want to *change* the behavior of the system. That is... you want a lot of your tests to fail, because you change the behavior. Now you not only need to change the "System under test", you also needs to change or rewrite your tests. This can be very hard to do without introducing bugs. Your tests will not make this change easier... on the contrary, that will make it harder.

8 How do we do do TDD when using products which generates a lot of code. E.g. when using Linq To SQL, Entity Framwork. Or if we "copy paste reuse" a lot of code form another project (we all do that, right?).
A: It is very hard to do "baby steps" when using code generation. I you try to do that, it will slow down you process significantly.

9 If we are new to TDD/Unit testing, how do we know we are doing it the right way?
A: Only time will tell. It is very easy to get it wrong. You risk spending a lot of time trying to introduce TDD/Unit Testing on the project. It requires a lot of discipline, form *all* developers on you team.

I'm relay looking forward to your book. I relay hope you will be able to show how to do it "the right way" without introducing a lot of nonsense code. As you can tell, I'm not at all convinced that TDD and unit testing is as valuable, as you (the community) wants everyone to believe.

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Jules — Sat, 24 May 2008 13:04:30 -0000

"1 and 4 are the same...."

Well, it _is_ a very important question.

"Tough question: where do I start unit testing existing code? Possible answers: write tests to fail for existing/known bugs, concentrate on modules with known quality issues (high bug rates, etc.), read Art of Unit Testing, etc... :-)"

The standard work on this subject is "Working Effectively With Legacy Code" by Michael Feathers.

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

willer — Tue, 20 May 2008 21:56:17 -0000

* 1 How much time will this add to the current process?
The limited data out there indicates dev takes 5-10% longer with TDD. However, the number of bugs that come out of dev are about 1/5 of the original. This is probably a 30% reduction in overall time.

* 2 Will my QA job be at risk because of this?
QA people are needed to script functional tests, write test plans, run exploratory (manual) tests, test scalability, scan for security holes, etc. QA is a much more integral part of an agile/TDD team ... their work is just more challenging and interesting than the traditional "here's a Word document with a table describing what buttons you will click" type of testing.

* 3 How do we know this is actually working?
The CI server shows green tests.

* 4 How much time will this add to the current process?
Same as 1.

* 5 Is there proof that unit testing helps?
Try it out, watch how much better the project feels with green tests. You'll see! There's no question in the community that it helps ... the difficulty is more with maintaining a sufficient level of discipline with the test building itself.

Re: ISerializable - Roy Osherove's Blog

shani — Tue, 20 May 2008 16:22:23 -0000

Hey, I have question that will probably jumb to every starter-tdder which is:
-- when my testing are practical, why? becuase there are many times (most of them at the begining) - while we wrote and spent a lot of time on writing tests. but we still have bugs, and we have the filling that the tests are not helping, or even worser we have our tests failing from time to time and we spend more time refactoring than coding the real code....

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Anthony Williams — Tue, 20 May 2008 08:12:33 -0000

1. Unit Testing will likely increase the length of time between first thinking about the problem and first having a piece of software that "sort of works". However, it will also probably reduce the length of time between that point and having a piece of software that really works in a robust fashion. If you're doing unit testing thoroughly, and really thinking through all the "things that could possibly break", the first cut might be the only one.

2. QA jobs are only at risk if all the QA team does is the manual equivalent of unit testing. However, a good QA team does much more than that: they exercise the product in interesting new ways, try out corner cases and evaluate overall usability as well as test the product. The QA team should work closely with the stakeholders and end users to ensure that the product will meet their needs in a way that users will be comfortable with. This can be much more rewarding than the drudgery of manual unit testing.

3. Bug reports from customers will reduce, and the QA team will find they have time to do the usability evaluations and so forth described in response to Q2 as the basic functionality "just works".

4. Same as 1

5. "Proof" is a tricky thing. Unit tests will help find bugs sooner. Finding bugs sooner saves money and reputation, as the bugs found by unit tests can be fixed before the product is delivered to customers. If fewer bugs are delivered to customers, support costs will be lower, as the customers won't have to report bugs so often, customers will be happier, and the reputation of the company is not tarnished by buggy software.

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Robz — Mon, 19 May 2008 23:55:23 -0000

Questions from other developers
1. So when you asking me to write tests, you are asking me to write more code. Why should I write tests when I can verify things with the debugger and with QAs?
Manual vs. Automatic: How much time do you spend in the debugger each time you are stepping through code? What if you forget to verify something? Do you restep through old code when you make changes to the code in some way? How do you remember how it's supposed to behave when revisiting in one year? How about 3 years? So if it takes you 2 minutes to write something, and you spend 30 seconds in the debugger each time, then the fifth time you would have to go to the debugger, your unit test would have paid for itself. And many times over the further you get in development. I run hundreds of tests when I make changes to code. In less than a few seconds I know if I broke something. How do you do that manually?

Another reason: In team development testing is the only way to have trust/confidence.

Re: ISerializable - Roy Osherove's Blog

John Aldrin — Mon, 19 May 2008 22:14:24 -0000

How do I get developers interested in unit testing.

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Jeffro — Mon, 19 May 2008 21:03:33 -0000

One thing I would love to see answered in your book is a common scenario that we come across in our shop quite often: how to unit test functionality that needs to process data in batches. Usually when we come across a scenario like this, the traditional way is to write a sproc in SQL Server to execute a series of "INSERT INTO x SELECT FROM y" statements, with pre & post processing around the statements. This method is more efficient when dealing with large amounts of data but is much tougher to write tests against, while pulling this logic into code would be less efficient (sometimes so inefficient that it wouldn't meet performance requirements) but is much harder to unit test. What's the best way to handle this scenario for testing purposes?

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Eli Lopian — Mon, 19 May 2008 14:11:17 -0000

You could use Ivonna: http://sm-art.biz/Ivonna.aspx

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Emanuele DelBono — Mon, 19 May 2008 13:01:08 -0000

I would add:
- How do I persuade the manager that wirte Unit test is a good thing?
- Can I write tests only for some portions of my code?

Re: ISerializable - Roy Osherove's Blog

Ofer — Mon, 19 May 2008 10:25:37 -0000

I can answer 4 if you give me the answer to 1.

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Sasha Goldshtein — Mon, 19 May 2008 10:17:03 -0000

- What do we do with the huge amount of legacy code that is not unit tested? Is it going to provide us any value if we only test the new code?

- How hard is it to switch the programmers' way of thinking into a unit test driven approach?

- Are we going to be coupled to a specific test framework now?

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Zack Jones — Mon, 19 May 2008 07:42:47 -0000

FYI - your web site is down with the following error:

DotNetNuke Upgrade Error
The Assembly Version ( [ASSEMBLYVERSION] ) does not match the Database Version ( [DATABASEVERSION] )

ERROR: Could not connect to database.

Cannot open database requested in login 'RoyoDNN'. Login fails. Login failed for user 'RoyoDNNAdmin'.

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Klaus Hebsgaard — Mon, 19 May 2008 07:30:11 -0000

6 Unit testing adds extra code to maintain
7 What if there is errors in the unit test code

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Heiko Hatzfeld — Mon, 19 May 2008 06:49:28 -0000

How much work is it to add unit testing my existing code?
- It can be applied to a ongoing project? (that is, one that started without unit testing in mind)

--> DONT do this. Retrofitting legacy code with unit tests is very hard. You should NOT pick this if you want to introduce unit tests into your organisation. Instead start with a "new" project or a new feature which can be seen very "seperated" from your main code. If you start with the "old" code, then you will have to deal with a lot of problems, and you will have a hard time to integrate your tests (After all, that code was not written with testability in mind). Build up skills, and after you feel confident, then switch to legacy code...

I have a mixture of code platforms (C++, VB6 and C#). Is unit testing feasable in this scenario?
-- Start testing the code that you can test... So stick with c# for starters, and try to isolate the dependencies for the the other components... But it depends on what calls what here ;)

> 1 How much time will this add to the current process?
Initially there will be an extra cost. But over time you will harvest the results. The sooner you start, the sooner you can harvest... Also, you have to worry about less legacy code. I think it is important to tell the management that it will initially ADD to the development time. If you dont tell them this, then they will see a slowdown, and start to rush you. Or make you stop writing tests. If this happens, then you are hosed... Its hard getting back from that point. So tell them the INITIAL Investment is there...

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

paulmair — Mon, 19 May 2008 06:15:26 -0000

How much work is it to add unit testing my existing code?
My new code needs to call into COM objects. Is this easy to unit test?
I have a mixture of code platforms (C++, VB6 and C#). Is unit testing feasable in this scenario?

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

jeffbrown — Mon, 19 May 2008 03:33:52 -0000

3. You know it's working when you begin to derive satisfaction from your unit testing effort.

This is often a problem for people who are new to unit testing. They are accustomed to the quick joy that comes from running their code in the debugger and spot-checking their latest work. Beginners to unit testing may perceive the benefits in their heads but fail to feel them in their hearts because gratification is delayed. In time, however, a new sense of pride should emerge: that derived of having a comprehensive suite of tests that proudly exclaims, in GREEN: "It works!" Then they can focus on other ways to improve quality instead of endlessly chasing after regressions in fear and confusion.

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

Hector Correa — Sun, 18 May 2008 22:37:33 -0000

> 1 How much time will this add to the current process?

How much? It depends. Really.

If you take baby steps and start simple (with only a few unit tests in a few places) it won't add much. But it definitively will add some time to your development efforts initially.

Something to consider is that a lot of development teams already write code to test their code, but they consider this code as throw away code. Formalizing this kind of code and keeping it alive does add a bit of complexity to the development effort, but as another of your readers indicated, it should be seen as an investment in the future.

> 2 Will my QA job be at risk because of this?

Not at all. In my experience QA gets excited about this because the quality of the build tends to stay high when unit tests are ran on a regular basis. This allows them to focus on important stuff rather than little details that now will get caught by the unit test.

> 3 How do we know this is actually working?
> 5 Is there proof that unit testing helps?

You'll notice that your builds are much more stable on a regular basis. You'll notice that refactoring is safer and that you can make significant changes to the system and know right away what will break rather than having to hope that you didn't forget to test something. They will not guarantee a bug-free system, but you'll notice a more stable system within a few cycles of doing unit tests.

[shameless plug]
I have a post on my blog where I touch on some of these questions
http://www.hectorcorrea.com...
[end of shameless plug]

Re: Tough Questions about unit testing - Your comments required - ISerializable - Roy Osherove's Blog

nsoonhui — Sun, 18 May 2008 21:59:20 -0000

I've written a post on this: