CSCI 5828 — Spring 2009

The Grades page has been updated one last time to reflect the grades of all CAETE students who completed project 4, project 5 & homework 6. I hope you all enjoyed the class and have a great summer!

The grades for Project 5 for the eight teams that presented today during lecture have now been posted. I will work on final grades for the class once I have received submissions from those CAETE students that were keeping to in-class deadlines. I will post another announcement here once I have submitted final grades to the registrar's office.

I really enjoyed today's presentations.... kudos for the high-quality work and sticking to the presentation schedule!

We have 8 teams that need to present and 75 minutes of lecture time. As a result, please keep your presentation to 10 minutes and be prepared to switch quickly between presentations. We will overrun our lecture time but if we act efficiently we can be done and be out of the studio before the next class starts at 3:30 PM.

To keep us on track, I'd like to follow this schedule:

Presentation 1: Flemming & Kooh

Presentation 2: Roberts, Mackey & White

Presentation 3: Allred & Knappe

Presentation 4: Khan, Pfeiffer, Yan & Zheng

Presentation 5: Beldyk, Gonzales, Gonzalez & Novinger

Presentation 6: Iyer, Noor, Philipose & Sundaravel

Presentation 7: Grosshans, Hammuda & Wichiencharoen

Presentation 8: Haller

In addition, I will present two signals during a talk: one at 8 minutes into a presentation that will indicate that a member of the next team should get up and move to the front of the stage; that will be a signal to the current presenters that they will need to wrap up their presentation. The second signal at roughly 9 minutes 30 seconds into a presentation that will indicate that its time to swap laptops for the next presentation. The current presenters can use those 30 seconds to answer questions while the next team sets-up.

Please work with me to ensure that tomorrow's session runs smoothly and that all teams get a chance to present!

I've updated the Grades page with scores for homework 5 and 6. If you missed points on Homework 5, the grader should be contacting you soon with feedback on how you lost those points.

Lecture 26 and Lecture 27 are now available on the Lectures page.

Homework 6 is now available on the Assignments page.

Slides for Lecture 25 are now available.

Slides for lecture 24 are now available.

The slides for lecture 23 are now available.

Homework 5 is now available.

Both the grader and a student in class pointed me to this testing tool:

Selenium: it is useful for performance testing a Web application.

Here are two articles discussing it: for and against

Another testing tool, this time focusing on Databases, is called dbunit. According to the student who sent me the pointer, it allows you to wrap unit tests in database transactions, but also to populate the database differently on a test by test basis and clean it up afterwards.

Just a quick heads-up: I am going to cover Chapter 8 from the concurrency textbook in Tuesday's lecture and will then cover Chapter 9 from Head First Software Development on Thursday. I will be posting details on Project 4 later tonight.

The slides for Lecture 22 are available on the Lectures page.

Slides for Lecture 20 and Lecture 21 are now available on the Lectures page.

I have a masters student who is working on a thesis involving the use of ambient user-interface techniques in agile software development. In particular, he has built a lighting system (see below) that displays the values of various software metrics. As the developers make changes to their system that impact the value of these metrics, the lights change color to reflect the change. My student has integrated his lighting system with the set of tools provided by RallyDev.

My student would like to meet with a few students in this class to demonstrate his system and get your reactions and feedback. If you would like to volunteer to meet with him (it would take no more than 30 minutes of your time), send me mail and I'll put you in touch with him. I'm hoping that at least a few of the students on teams that are using RallyDev's tools for the semester project will be willing to meet with him. Thanks.

I have now finished grading Project 2 and have updated the Grades page in response. I will now turn my attention to finishing the HW 2 re-grade.

Of course, the best way to deal with a deadlock occurring in your Java program is to change your FSP to model the situation of multiple threads accessing the account and ensure that LTSA declares your model deadlock free. Then translate the model into Java and your Java program should be deadlock free.

Hint: if you're Java program is deadlocking on the conditions mentioned in my previous What's New post then you are not handling mutual exclusion on the account correctly. In particular, you don't a thread making its choice about whether to do a withdrawal or deposit before it knows what the balance is. What needs to be atomic in the system is learning the balance and then making a choice about whether you want to do a deposit or a withdrawal.

I have updated the Grades page with information on the received/not received status of review 6 and project 2.

If you haven't sent in your team evaluations for Project 2, please do so now. "Teams" consisting of only one member do not need to send in an evaluation. :-)

I have regraded roughly 13 of the Homework 2 assignments and sent updated scores to the relevant students. I will regrade the remainder of those assignments during Spring Break and will send updated scores to students as they become available.

Example Code that I demonstrated on Thursday for the Pot of Soup problem

In reviewing the Project 1 assignments, a few other tools surfaced that can be used for coordinating small teams:

Unfuddle: http://unfuddle.com/

Fogbugz: http://www.fogcreek.com/FogBugz/

Jude Allred has informed me that Fogbugz is free for student projects. Simply create an account and then send them an e-mail indicating that your account is being used to manage a project for this class.

I have officially updated the due dates for the Project 2 assignment, as discussed yesterday during lecture.

FYI: Google will not allow me to post the slides for Jim McMaster's talk yesterday. Jim sent me a few pointers that point to additional material on Test-Driven Design.

First: Dave Astels book: http://www.amazon.com/Test-Driven-Development-Practical-Guide-Coad/dp/0131016490

I was able to process the individual grades quicker than I thought. The grades page now shows the results for those people who took the midterm yesterday. If there is a "0" in your row for the midterm that simply means "not received". I will discuss the midterm tomorrow in lecture and will review the grading for homework 2 on Friday.

Hi. One of the questions on the midterm has the following two sentences in it:

… List two problems that it can help us to avoid in software development. If it’s easier, you can also state your answers in terms of the benefits this principle provides.

To avoid confusion on the test tomorrow, what these sentences imply is that you can either:

• list two problems
• OR list two benefits
• OR list one problem and one benefit

Now all you have to do is figure out what principle I'm talking about! ☺

Take a look at this article that points at some classic papers from the SE literature:

10 Papers Every Programmer Should Read

Enjoy!

Hi. I will be holding office hours tomorrow in ECCS 111 to allow students to come and ask questions concerning the midterm from 9 AM to 10:45 AM and from 12 PM to 1 PM. If you can't stop by then send me questions via e-mail and I will do my best to answer them in a timely fashion. Note: please come with specific questions to make the time productive and note that I will NOT be answering questions concerning the re-grading of HW 2, as I won't be able to spend any time on that issue until Friday, March 6th. Thanks!

On Thursday, I presented a short example to demonstrate the effects of alphabet extension on the composition of two FSP processes. A student asked me to post that FSP. Here it is:

range T = 0..50
set VarCoins = {in.coin[T], out.coin[T]}
COINTRACKER = (in.coin[5] -> out.coin[10] -> COINTRACKER).
TROUBLEMAKER = (in.coin[5] -> CHOICE),
CHOICE = (out.coin[10] -> TROUBLEMAKER |
          out.coin[15] -> TROUBLEMAKER).
||TRACKME = (COINTRACKER || TROUBLEMAKER).
COINTRACKER2 = (in.coin[5] -> out.coin[10] -> COINTRACKER2) + VarCoins.
||TRACKME2 = (COINTRACKER2 || TROUBLEMAKER).

Okay, thanks to the hard work of the grader, I now have all the grades for Homework 2 posted. We will discuss that assignment in lecture tomorrow.

The previous entry contains a pointer to the script I used today to give the subversion demo in lecture. Feel free to use it to try out subversion on your own machine if you have little or no experience working with subversion. You will need to modify the start of the file:/// URLs but should otherwise be able to use the script as is.

Today, in Lecture 13, I mentioned Microsoft's Visual Source Safe as being a configuration management solution for the Windows platform. I received a message from one of our students (who works for Microsoft) disabusing me of that notion:

Noticed you had Source Safe as the Microsoft source control product. Source Safe hasn't really been used as a source control for Microsoft for a while now. I don't think it's even supported. TFS is the primary system (specifically Team Foundation Version Control) for source control now. Microsoft product teams use 2 main source control systems. TFS mentioned above and an internally developed system called Source Depot.

If you ask anybody familiar with Source Safe, they cringe...including me :)

Thanks for the correction!

IEEE Software, a top SE journal, has released its picks for the top 25 papers that have appeared in its pages over its 25 years of publication. Three of the papers are available for free for all comers until March 31st. However, as students, as long as you are logged into the campus VPN, you should be able to access all of these papers in the IEEE digital library. Enjoy!

Lecture 12 is now available on the Lectures page.

ST-Ericsson and ARM have announced that they will be demonstrating the world's first symmetric multiprocessing-enabled mobile platform this week at the Mobile World Congress in Barcelona. Multi-core is everywhere! :-)

One of our students forwarded me an article on Google's use of "beta" applications and its process for deciding when to "pull the plug" on a project. He thought it complements the discussion of "good enough" vs. "perfect" design as covered in Chapter 5 of our Head First Software Development textbook.

The article is located here:

How Google Decides to Pull the Plug

Key Quote:

Perfection closes off the process,” Mr. Jarvis said. “It makes you deaf. Google purposefully puts out imperfect and unfinished products and says: ‘Help us finish them. What do you think of them?’

Lecture 11 is now available on the Lectures page.

I discovered a problem with Lecture 9 and so uploaded it to the website again. And, I have added the slides to Lecture 10 as well. Both can be found on the Lectures page as usual.

Interesting discussion on Slashdot about dealing with issues of scale: Facebook currently has 140 million users and has been adding 600,000 new users a day. The article points to a post by the Facebook engineering team on things they've been doing to keep up with millions of users pounding their systems.

In another example of how software engineering is concerned with issues of scale, check out this article that inventories the infrastructure needed for large-scale web sites. Here's an additional blog post from a developer who works on such a site about some of the tradeoffs that occur in achieving such infrastructure.

In lecture, I've said several times that I wasn't aware of work being performed to scale agile techniques to large development organizations. This assertion was too tempting of a target for the universe to ignore and so yesterday I received in the mail a new book by Craig Larman and Bas Vodde entitled:

Scaling Lean & Agile Development: Thinking and Organizational Tools for Large-Scale Scrum

Click the link to check it out. I'm glad I was wrong!

I'm finding it difficult to include material on traditional software engineering estimation techniques while also covering the material from our software development textbook. As such, I'd like to direct students interested in this topic to the slides from Lectures 9 and 10 of the Spring 2008 version of this class. In particular, slides 41 to 65 contain pointers to a variety of estimation techniques and you can use the author names associated with those techniques to look up more information on each technique. This information is NOT required for this instance of 5828. FYI.

Review 4 and Homework 2 are now available on the Assignments page. Review 4 is due by midnight on Thursday of next week and Homework 2 is due by Monday of the week after next. (In other words, I'm giving you two weekends to work on Homework 2.)

I received responses from both Jeff Magee and Jeff Kramer concerning the question raised in the previous entry. Both agreed that the two specifications are semantically identical and that the non-optimized LTS can be converted to the optimized version by asking LTSA to minimize the non-optimized version. To see this for yourself, fire up LTSA and enter the first FSP specification:

const N = 1
SUM = (in[a:0..N][b:0..N] -> out[a+b] -> SUM).

As we saw last time, this spec produces this LTS:

Now click the minimize button, as shown below.

LTSA will minimize the LTS and it will now look exactly like the LTS produced by this specification (see previous entry for details):

const N = 1
SUM = (in[a:0..N][b:0..N] -> TOTAL[a+b]),
TOTAL[s:0..2*N] = (out[s] -> SUM).

Now for the responses. Here is what Jeff Magee had to say:

Hi Ken,

The compiler does not guarantee to produce minimal LTSs. In the second case as the states are declared explicitly, there is only one state generated for both 0+1 and 1+0. In the first case, a state for each state is implicitly generated. Minimizing the first LTS will return the second, so as you note, they are exactly equivalent.

Cheers

Jeff

And, here is what Jeff Kramer said in his reply:

Hi Ken,

This is just the way that the FSP is compiled and in the first case, without knowledge of the semantics of +, a+b might not be the same as b+a, whereas in the second case the state is first computed and becomes s.

The first case can simply be minimised to produce the second as they are observationally equivalent.

Best wishes

jeff

In interpreting their response, the second FSP is explicitly defining a destination state as represented by the TOTAL process. So, LTSA computes the explicit state that will be reached when control transfers to TOTAL and is able to determine at compile time that the in[0][1] and in[1][0] actions both go to the same state. In the first specification, it makes no assumptions about the semantics of the addition operation and so conservatively generates different states for in[0][1] and in[1][0] leading to the non-optimal LTS that we saw above. Once we invoke the minimize operation, it then looks at the state machine more closely and realizes that two of the states in the non-optimal LTS have the same semantic meaning and collapses those states into a single state producing the optimzed (or rather, minimal) LTS.

Hope this helps and thanks very much to the student who asked the question that led to this discussion!

Review 3 is now available at the Assignments page.

Lecture 6 is now available on the Lectures page.

I have added a Grades page to the website showing the grades for Review 1. The table on that page is indexed by a mysterious attribute called Id.

To calculate your Id for this class, use the following procedure.

1. Start with your CU student id, for example "999-99-0234".
2. Take the last four digits, e.g. "0234".
3. Convert to an integer and think of that number as tmp, e.g. 234
4. Take your last name and make it a lower case string, for example, "anderson".
5. Convert each letter of your name into a number, a == 1, b == 2, … z == 26.
6. Sum up the numbers of your name and think of that number as sum, for example, the sum of "anderson" is: 90.
7. Add sum and tmp together to compute your Id for this class.

If you do this calculation and do not find your Id in the grades table, contact me and we'll make sure that we both used the same input into the algorithm.

Lecture 4 is now available on the Lectures page.

I fell down on the task of creating review 2 yesterday. As a result, I've decided to release it tomorrow and will push its due date back into next week. Stay tuned!