Bug in uniq on my system
shlomi:~# cat uniq-test.txt uses/bio-info uses uses/qa shlomi:~# /usr/bin/uniq < uniq-test.txt uses/bio-info uses/qa
A bug isn't it? My system is Mandrake 9.1, and neither the uniq from RedHat 6.2 nor the latest from the GNU diffutils exhibits this problem. It broke one of my scripts and so I installed the latest uniq to my $HOME/bin directory.
Objectivism and Open Source Essay
I wrote an essay titled "Objectivism and Open Source" today. It aims to explain to Objectivists why open source is a Good Thing<tm>. Read and flame away!
Subversion Issue #854 Revisited
Apparently, I did not notice some comments in the issue in the Subversion bug-tracker, that said that the behaviour has changed, but still was not as expected. So, the bug was re-opened and I created a new fix, that checks if "svn revert ." is done on a newly added directory and throws an error if it does. Quite a trivial fix, but I also wrote a regression test to make sure this is what happens.
It was eventually applied with small fixes. One less bug before Subversion 1.0.
Larry McVoy Interview
I went to the BitKeeper homepage today to recall the name of the push command, and saw on its front page that Linux World conducted an interview with Larry McVoy. This was partially in response to a previous interview with Bruce Perens. I wonder how I could have missed it.
I searched the Perens interview for BitKeeper and read this part. Basically, Perens accuses McVoy of threatening him every time he meets him. McVoy denies these accusations in his interview. It's one's word against the other. I do know Larry McVoy has threatened several other people with ligitation online. And he does seem to me to be quite hot-headed. So I tend to believe Perens.
This all brings me back to my conclusion that reverting the BitKeeper license would be a good idea.. The reason the source code was hidden in the first place, was because some companies modified it to subvert BitKeeper and disable the OpenLogging facility. This is all very well, but I doubt a substantial number of them would have become paying customers.
With its current licensing scheme, BitKeeper would never become the state of the art. Never. While it is technically superior to any free software alternative (at least for now), hardly anyone dares touch it with a ten-foot pole because of its license. If its licensing change to something much saner, and its source code be revealed, it could, OTOH, eventually conquer the market and be distributed as part of major distributions. I don't know why McVoy does not realize that, but it is mainly his loss.