Dear Software Vendors October 2, 2008
Posted by Kurt in : Computers , add a commentPlease do not make it hard for me to make your software available to my faculty and students.
As awful as I find the whole flexlm license server thing to be, it is much better than other methods. Dongles are bad for client systems. In a university, you are just asking for the helpful little bit to be stolen. And if some idiot suggests I glue the dongle in, then your software will just never make it onto my systems.
What brings this on? This quarter, faculty requested getting some new software on the lab image. No problem, I think. It installed fine. However, the licensing is butt stupid. It is MAC address locked. I know it is considered clever, but it’s not. MAC addresses really aren’t immutable. And what it really means, is that my normal proces of cloning machines doesn’t work for this software. So I can clone it in a non-functional state. But to make it work, I have to gather MAC addresses for 80-something lab machines (and no, not enough licenses for all of our machines anyway), and get 80-something licenses issued. I then have to wrangle 80-something licenses onto each machine individually (And that’s assuming when I get these licenses issued, I don’t have to do them one at a time, which I am afraid I will have to do so, instead of some sort of sane thing where I can paste them all in at once).
Seriously, system administrators work hard, and are the key to getting your software out to the users. ESPECIALLY in an academic environment. Why would you make it hard for me? Because someone might steal it? I’m sorry. I know you think your software is super-special in the area it addresses, but really, I don’t think that my students will bother stealing it. And honestly, if they did somehow, the students in my department are software engineers, going to all sorts of influential companies. Oh no! One lost sale! Except our graduates are more likely to get your software used in the real world. So, to keep a student from having the software (who was *not* going to shell out money on your tool, let’s be honest), you’ve sacrificed many potential sales at their big future employer. Good job.
Contributing back to OpenBSD November 2, 2007
Posted by Kurt in : Computers , 1 comment so farSo today, I finished making a patch to OpenBSD that allows the serial card I bought to work properly.
Basically, it was a new version of the card by this vendor. Except they changed how it worked. Really, this card is a PCI bridge with 2 4-port serial “cards” behind it. The existing driver recognized the chips used, however, the card didn’t work. It took a bit of investigation to find out that some newer cards (like this one) use a timing crystal that is ten times faster than what is usually used. So the existing driver would put garbage out the serial port. (I’m not sure if it was basically running at 96000 baud instead of 9600, or more confused). Easy way to deal with that, it’s just that the driver didn’t know this revision needed that tweak, because it just saw it as generic versions of this card.
Into the world of PCI I went. I ended up needing to get the PCI vendor and subsystem numbers using lspci. Which isn’t in a base OpenBSD install by default. And it’s not on the 4.2 install media I bought (only the most useful subset makes it onto the CD). And since this was before 4.2’s official release, no package available for download. Then I had some fun with getting it to compile, since apparently the expat library moved into the base x packages instead of being a port, but the ports tree didn’t realize this. So I got that resolved. I get my numbers.
Not I dive into the kernel source code of OpenBSD. I found that I needed to change the pcidevs and pucdata.c files. But how? It took a bit of bashing my head against the code, and some mistakes that caused the machine in question to just freeze, but I got it going.
So today, after that, I created diffs for those files and submitted them to the appropriate OpenBSD mailing list. Just got back to my desk a few minutes ago, and there was a response to the list from one of the main developers saying it was a bit garbled, and he had to apply it manually, but that my contribution has been committed.
So I’m feeling all proud because I have submitted code to my favorite software project and had it accepted. It’s really not like it’s a big contribution, but I actually took the effort to learn and do it myself.