Abram Hindle's Blog (circa 1993)

Abram Hindle's Blog (circa 1993)

Thu, 06 Sep 2007

Subject: Corrupt countries were more likely to support the OOXML document format? With a correlation coeffecient of -0.31?
http://www.effi.org/blog/kai-2...

For some reason they use a Wilcoxian test and fisher t test on the data. The data has 2 random variables, corruption (continuous) and vote (yes/no)

They said they should with good p-values that there correlation between voting for approval and corruption. Well using both Pearson and Point-Biserial Correlation coeffecient I found a coeffecient of -0.3129 with a p-value of 0.009373 this means there was a weak or medium stength linear correlation between corruption (more negative) and approval votes.

What's interesting is that EFFI keeps applying the wrong tests when they want to show to correlation. I also did the rank based correlations and they weren't as good as the linear cases -0.29 for Spearman-Rho and 0.24

 for Kendall-Tau rank based correlations. So there is a correlation but

not a really great one, nothing to write home about.

My data and a log of what I did is at: http://churchturing.org/w/ooxm...

abram

[] permanent link

Mon, 30 Apr 2007

Subject: How to stop Digg Cheating Forever
http://slashdot.org/articles/0...

[] permanent link

Thu, 08 Feb 2007

Subject: Blurring
http://dheera.net/projects/blu...

I wouldn't have gone about this problem the way he did. I would've used feature vectors and machine learning to characterize and classify the bitmaps (which he'll probably have to do). Though what I thought was great is he actually analyzed the problem and came up with a very novel solution (I doubt it works in all cases but still close enough).

[] permanent link

Subject: Iannis Xenakis Composition Tool
http://bradfuller.com/blog/200...

[] permanent link

Tue, 31 Oct 2006

Subject: Music Reactable
http://www.youtube.com/watch?v... This is so amazing it makes me jealous.
[] permanent link

Mon, 09 Oct 2006

Subject: Web 2.0 Security
A top 10 of web2.0 security risks. http://www.net-security.org/ar...”9
[] permanent link

Subject: Automated Testing
Automated Testing of Open-Source Music Software with Open Sound World and Open Sound Control

http://www.opensoundcontrol.or...

Using open sound control to test audio software. Good idea except I can't read the paper from that link :(

A marginally related paper: Automated test case generation for the stress testing of multimedia systems http://www3.interscience.wiley...
[] permanent link

Mon, 18 Sep 2006

Subject: Promotion of FUD
http://crave.cnet.co.uk/0,3902... http://slashdot.org/articles/0...

These are examples of how Youtube gets used by groups with an interest in the promotion of false information.
[] permanent link

Sun, 17 Sep 2006

Subject: New Use for the Research Blog
Alas I rarely update the research blog.

I think I will now it use to store bookmarks related to software evolution and other interests.

Confessions of a Recovering NetBSD Zealot

http://www.onlamp.com/pub/a/bs...

The important part of this interview related to my work is the promotion of commit policies as good software development practice. Especially for open source where if you commit unfinished code it ruins compiles for many people.

I think commit policies are great for the development tree but I also think it highlights a failure of CVS. People use CVS to record revisions and act like a backup system. If people want to save their work for the day but have to follow a commit policy they can't commit their changes to the central repository. Sure they could use a branch but if you've ever used CVS you'll know that branching and merging are not fun at all. Thus an approach such as darcs or some versions of svk where you can have your local repository and then push all your changes when you are done is possibly better. It allows you to use version control on patches which are working on and then allows you to push these changes once you are done.

Regardless commit policies seem like a logical rule to enforce some semblence of software quality or even promote requirements such "the software always runs". This is an important quality for OAS in particular.
[] permanent link

Tue, 08 Aug 2006

Subject: Java, Subject Verb Object notation
My friend mikael forwarded me this blog entry:

http://steve-yegge.blogspot.co...

I don't like this. He ends up arguing for tool support in java for functions but can't see that classes are already used in java as functions which can be passed around. It is like he's making a weak argument for functional programming but doesn't realize that functional programming includes passing functions around (which can be like passing an object around when all you do is plan to call upon a method it implements).

My friend mikael suggested he is complaining about the Java community creating code like this. Mikael doesn't have the experience in the area for me to trust but I don't have the experience either.

Is real java code like this? Or do people often use classes just to hold functions?

abram
[] permanent link

Wed, 21 Jun 2006

Subject: Desktop Metaphor
http://www.youtube.com/watch?v... It is a shame they waste so much time on punting icons around when the idea of managing programs and documents together sounds so much more interesting (especially to those of us who don't use a desktop).
[] permanent link

Wed, 24 May 2006

Subject: Long time no update
I haven't updated this is blog in a very long time.

So far I've published:

Abram Hindle and Daniel M. German: SCQL: A formal model and a query language for source control repositories , 2nd International Workshop on Mining Software Repositories (MSR 2005), May 2005, 100--105

Daniel M. German, Abram Hindle: Measuring Fine-Grained Change in Software: Towards Modification-Aware Change Metrics. IEEE METRICS 2005: 28

Daniel M. German, Abram Hindle: Norman Jordan: Visualizing the evolution of software using softChange. SEKE 2004: 336-341

And I have a masters thesis, You can always email me for that.

I also got a Masters in Computer Science from UVic under the great supervision of Dr. Daniel M. German.

I am now studying at the University of Waterloo under Dr. Ric Holt and Dr. Michael Godfrey.
[] permanent link

Subject: Research Home Page
http://swag.uwaterloo.ca/~ahin... Here is my research homepage at U o Waterloo.
[] permanent link

A company has implemented it! We implemented a prototype but these guys have been more thorough and done it well :( WHY DIDNT WE PUBLISH :(
[] permanent link

Tue, 14 Sep 2004

Subject: What do you have to do?
1. Finish that paper 2. Get your CV done by Friday (include all the volunteer experience, the CSCU etc, coding competitions, perl mongers, open source stuff, games club (look at your presentations). TA for XP. etc.

abram
[] permanent link

Sat, 21 Aug 2004

Subject: Clustering
A good clustering overview.

http://genome.imim.es/~eblanco...
[] permanent link

Fri, 20 Aug 2004

Subject: Music Similarity
I've been using George Tzanitakis's software package Marsyas for analyzing my music collection. Using his extractor I extract a feature vector from each music file and then take those feature vectors and cluster them.

http://marsyas.sourceforge.net...

I couldn't get George's clustering tool to work so I found a useful library for perl Algorithm::Cluster:

http://bonsai.ims.u-tokyo.ac.j... http://search.cpan.org/~mdehoo...

Anyways I was easily able to use this handy software package to cluster the data extract by Marsyas.

It turns out that the clustering seems to work a little bit. It's hard to tell right now because the part of my music collection it extracted has been many noise. It was neat to see my music associated with other artists like the Haters or Merzbow.
[] permanent link

Thu, 12 Aug 2004

Subject: Paul Graham, McCarthy, Perl Python LISP
In my annotated Bibliography of the Early LISP History (1956--1959), I mentioned the reason for eval in lisp was for purely theoretical reasons. http://funnt.ca/~funnt/bib/#St...

Well it was interesting today as I read the slashdot post http://developers.slashdot.org... The Python Paradox, by Paul Graham http://www.paulgraham.com/pypa... . It talks about thr reasons people learn python. Less for money more for programming. I'm interested as I don't really care about Python that much I've done enough research to figure out it maps pretty closely to perl and that must be reason people have made perl parsers which parse python but run it in perl. Anyways he linked to his popular article Revenge Of Nerds http://www.paulgraham.com/icad... which started talking about the lisp eval function.

I like these Coincidences.
[] permanent link

Tue, 10 Aug 2004

Subject: ACM Ethics
I noticed on the ACM website they have a http://www.acm.org/constitutio... Code Of Ethics for all their members.

In this code of ethics they suggest you are to respect IP and that includes Patents. Software patents are included. I think it is really silly that as a member of the ACM we are expected to respect the ludricious claims of so many software patents holders, that they somehow own instructions you came up with. Even worse many of these patents are intentionally vague and thus can be applied to too many. Like the Eloas issue with Microsoft.

In Memoirs of software pioneer by M. Goetz, we learn initally software patents were used by the small companies to fight off the larger companies (IBM). Now it seems muddled with corporations assuming more and more power and simply forming agreements they wont sue each other. I can't say I agree with Goetz's purpose in trying to get software patents recognized, it seems to me more like a company trying to compete using the legal system instead of a good product. Of course the problems at the time were related to IBM bundling software with their machinery and giving it away. Which in late 60s and 70s seemed like a really bad thing but now the world has changed..
[] permanent link

Subject: Computer Virus Class
Saw a lecture about a prof at the University of Calgary running a course on computer viruses:

  Last fall the University of Calgary offered a course on
  computer
  viruses and malicious software, the first of its kind in Canada and
  one of only a handful that have ever been offered worldwide. The
  course generated a global stream of controversy and media attention,
  in part because students were taught how to write viruses as well as
  defend against them.

  I will talk about the rationale behind the course, what I taught the
  students, how we put together a secure laboratory for course work,
  and the experience we gained.

It was interesting. They made a special lab and tonnes of security so no code could escape. Students even couldn't download copies of their assignments. He didn't seem very interested in the fact that the students worked hard and paid money but couldn't keep property which was rightfully theirs. I really could care less what a document or software does, if you made it, it is yours (or at least you should take a copy of it for your own use (there are issues when you work for a company)).
[] permanent link