Monday, January 26, 2009

Has Git actually won?

I think one of the things hampering the world moving on from svn has been the lack of a clear successor. Both hg and git (and bzr earlier on) were getting a lot of attention which means that neither was an obvious choice. But git has been getting the air time recently. I've toyed with both and have to admit a preference to git myself, too.

The git port to windows (msysGit) has also been good.

I guess if we ever see Google Code hosting adding git support, that would be a real sign. Curious to see. (For a few objective and also some personal reasons, I'd rather use Google than GitHub, but not everyone will have the same concerns I do.)

In any case, I also find using git for local history tracking to be a nice choice. Not quite so automatic as Eclipse local history, but it allows more definite control and beats some other options.

Does anyone else have an opinion? Has git beat hg for mindshare as the successor to svn?

Friday, January 23, 2009

Are killing and government control inescapable?

My second political post here. Apologies in advance, but do we have to trade war-mongering (and big government) for baby killing (and big government)?

Maybe I'm grossly misrepresenting everyone, but I'd love to find politicians to vote for that are honest, charitable, and competent. I know that means playing tough sometimes, but I think a lot of caution is in order for certain topics. Oh, and knowing how to balance a budget would be just sweet.

And, yes, I'm in the US. I can speak in an even less informed fashion about politics elsewhere.

Monday, January 19, 2009

Maybe Blender for Robot Simulations?

In the recent past, I've been eying JBullet and/or jMonkeyEngine and jME Physics for robot simulation. I think these could work, but I'm starting to like the idea of Blender instead. Blender is an opaque interface for newbies like me but still a lower barrier to entry for so many features than what I find trying to learn jME. At least for me right now. And I can do Python, too.

One concern I have with Blender is the GPL license, but that's not an immediate concern. I do love that I can download a binary easily on different platforms. And it might be interesting to make a network (or even web) server driven system with it, anyway. Easy way to allow for different languages, too.

Here's a nice detailed investigation into the use of Blender for robotics, by the way.

(Oh, and if you've heard of it, I might like Gazebo more if they distributed binaries and/or didn't try to make it so hard to build. Talk about having a billion dependencies on the latest version of everything.)

Monday, January 12, 2009

Use amsmath, not eqnarray

Being bothered in LaTeX by my difficulties with eqnarray, I searched a bit and found the recommendation to use amsmath instead, which generally already comes built in. I guess I'll give it a try.

LaTeX sure is a complex ecosystem. I like to imagine having better alternatives, but I haven't yet seen any. Well, someday, I may try to stick to pure web environments for most cases. Do I really care about ensuring sweet ligatures and so on? We read tons of web pages all the time. There've got to be tools for LaTeX-to-MathML (or images), but for actual conference and journal papers, I have a hard time imagining ever switching from LaTeX. Though, really, I'd love to see a conference or journal emphasize HTML publishing over PDF someday. I think it's possible.

Just that knowing HTML well and knowing LaTeX/PDF well seems like a lot of overhead. I'd like to limit the number of things I need expertise in. Maybe someday.

Wednesday, January 7, 2009

haXe to C++ Compiler

Um, compiling haXe to C++ (in addition to SWF, JS, and PHP) just might make it a killer platform. Maybe I should invest some time in it.

Tuesday, January 6, 2009

Java Open Sourced and Abandoned?

I'm tempted to read "instead of producing JDK7 we did JDK6u10 and JavaFX" as "Sun open sourced Java SE and then abandoned it to work on proprietary products (with some pure GPL side effects)". I think that's a bit too cruel, but it's also somewhat true.

The interesting thing is that few people have cared too much. I guess most Java is used in enterprise settings where stability is favored over bleeding edge. The Java community at large is very conservative when it comes to updates and changes. And we're used to just going along with Sun, too, I think (for core Java SE, at least). I think many other projects out there would have been forked by now if something like this happened.

Anyway, hopefully things get back on track some day (with nice licensing for the plug-in, JavaFX, and so on and/or progress on OpenJDK). I mean, Flash/Flex is about as open as JavaFX right now, if I understand things correctly. (And if you really want open, watch HTML 5, WebKit, Mozilla, and such.) Why not just go to the market leader?

Friday, January 2, 2009

Except that Java has the tools

It's fine to say that other JVM languages have closures (or whatever), but those other languages don't have tools like Java. Is it a sin to ask for an open source language with good tools and good performance that's actually a good language, too? Maybe some day. It doesn't have to be Java, but no one is in that sweet spot right now. And for some folks, getting the right features in Java seems the shortest path.

Lua: A Better JavaScript?

It seems I've been in "investigate every buzzword language on the planet" mood recently. Now I've done a quick glance at Lua. Seems to be a LOT like JavaScript, though different in many subtle ways. Some ways it's better is its sweet implementation (tiny size, tail recursion, "stackless", high performance), but it also seems to have easier namespacing and goodies like operator overloading. If only all the browsers would just integrate it, ... Or if they are as close as they seem, maybe a Lua to JS compiler would be straightforward? In any case, again, it doesn't seem a dramatically different language.