I am just now jumping on another "opportunity" to do the same. 
Thanks for the kind reminder LS!  Just because I am old, does not
mean I can learn Android OS programming :-)

Perhaps you should huff a lot of nitrous oxide. It doesn't help with the programming--quite the opposite. But it may help with the hubris.

Oh, maybe I'm only saying that because I'm obsessed with dentistry of late. But programming IS a lot like pulling teeth.

Discovered Project Euler (https://projecteuler.net/) for after one figures out FizzBang (http://blog.codinghorror.com/why-cant-programmers-program/).


Another good reference: http://docs.python-guide.org/en/latest/

I had read that article in blog posts at work discussing the issue perhaps a year ago, understanding the gist without getting why that ought to be so trivial. I happened to re-read one of those posts again after I was about halfway through the codecademy.com stream, and I found myself putting it together in my head. My python solution probably isn't particularly fast, but it does fit in a Twitter post, so there's that. Project Euler is full of fun, wacky questions to try to solve along similar lines, although I could wish for something more general problem-solving and less math-specific.

Good programming stands on a mountain of failure.  A good programmer
is someone who has failed in just about every way imaginable, but
didn't give up.

A great programmer is someone who continues to fail in just about every way imaginable, but ships the product anyway.

Only if he's pointyheard. Well, I guess that excludes Aahz these days..

This isn't strictly a programming thing, but it's at least related to it, so...

I need to build libffi using VC++ 2013, because I need to build gtk+ for Windows.

This site purports to have a spiffy-easy way to do that:

http://hexchat.github.io/gtk-win32/

And initially it really does seem that easy, until you get to libffi.

That library will not ./configure for me at all.

It eventually stops (without an error message), not having created a make file. I kinda need the make file for this library in order to build it.

I tried following the directions that he followed, here:

https://wiki.gnome.org/action/show/Projects/GTK+/Win32/MSVCCompilationOfGTKSta ck?action=show&redirect=GTK%2B%2FWin32%2FMSVCCompilationOfGTKStack

But they do not work. I get several crashes, a ton of messages indicating that I have permissions issue (likely between the Windows shell and the msys bash shell), and I see configure make some surprising observations, like how a double requires 0 bytes, and a long double only requires 0 bytes as well.

But at the end of the day (and yesterday, this was a literal statement), I can't get a Makefile out of ./configure.

I suspect from all the terrible and dire error messages that there's a problem between msys's bash and the command line. I'm on a 64-bit Win7 machine, so I'm having to make sure I use the 32-bit command line (or running bash is a completely non-starter). If I run bash without --login -i, it seems to miss several features I need for ./configure to work... perhaps because the paths are screwed up. If I run it with --login -i, I get the funky error messages, unless I run it without having run the cmd.exe first... in which case I can't get all the environment stuff I need for VC++.

Grr.

It did.

I think I've figured out that something is very broken in my environment.
The toolchain works in a virtual machine that we have, so I'll build it there.

But I ought to figure out why my environment is so borked. I suspect it might have something to do with the weird crap I'm doing in the course of testing things that leave my system in a funkier state than I thought.

I started with trying to make my ipython notebook accept interactive input.

Turned out wheezy had installed a really OLD version of ipython.

So then I tried to get a newer version.

I succeeded. It was shiny. And also python2 only, which is lovely, but I am writing mostly python3 and don't feel like rewriting things; I should be able to use either.

More work. More reading. More Googling. More not finding what I need.

An hour later, my poor Beaglebone Black is transitioning to "testing." We could be here a while.

NOthing I'm doing requires the latest and greatest; it's certainly not worth learning a new distro over.

It was hairy. For a while, I didn't even have /sbin/shutdown; not sure how I did that. Anyway, eventually, computer was stable again and all is well.
I have ipython able to do 2 or 3 at will. Everything else still works.

New to me, of course. I have dabbled in quite a few, but never that one.

slackware - there's a blast from the past. it's still around? i downloaded several floppies worth over 14.4kbps modem.

If you intend to distribute your code, you *shouldn't* be depending on the latest and greatest.

Holy fuck...

Working with managed code (whether Java or C#) puts me in mind of Apple creating a programming language or something.

Like an Apple product, you're expected to just use it the way they want you to use it, to do the limited things they want you to do with them, and just shut up about any complaints.

If you want to do anything even slightly interesting, it feels like they gently (at first) tap you on the hand, wag a finger, and say, "no no no... try again." But they don't bother to explain why they think you're wrong.

I expect you can find legions of software engineers lost in the labyrinth of managed codelets, swimming in a broth of warm code that doesn't quite do what you want, but gets close enough to drive you insane. And they're probably all completely bonkers by now.