Root Access

Shiki

19:00

I want to get a "gamer machine" (17 inch ASUS) because of the proper cooling

Uhm... real hard glitches. Even from using Chrome or the 3D acceleration in Flash.

allquixotic

@Misha Starcraft 1? Windows?

Misha

@al

@allquixotic Yes

@allquixotic all Populous 3

allquixotic

Their Windows drivers are completely and utterly separate from their open source drivers, because they are designed by different teams and don't share knowledge or code.

Shiki

There is a tool for Starcraft 1 on Windows.

Oh I have the bugs on Windows. (But they happen on Linux too.)

Misha

The tool disables the hd acceleration, sending it through the directx reference driver :-)

Shiki

19:01

It's definitely a hardware problem.. on both laptops

allquixotic

No, the rendering glitches are not hardware problems, at least not on Linux.

Sigh it comes down to the fact that the OpenGL and Direct3D APIs are not rigorously specified.

The APIs, basically, have numerous areas where the spec is open to interpretation.

Nvidia interprets it one way, AMD interprets it another way, and Intel is left somewhere in the middle.

What happens is that the application developers try to determine if you're using an Nvidia card, and if so, code it one way; and if you're using AMD, code it another way.

Shiki

That's why they should code everything in Java.

(lol.)

allquixotic

But many of them don't include any code at all for special paths to cater to the Intel drivers, so Intel ends up trying to emulate what one or the other vendor did, but it's very easy for them to get caught up in a bad place because of the way the apps are coded to behave differently for different hardware.

sidran32

@allquixotic Actually, that's not because of the DX or OGL specs.

That's because the different chips have different underlying architectures, and so certain styles of coding will perform differently on the different architectures.

allquixotic

@sidran32 Isn't it? The specs don't give an exhaustive list of valid function interleavings, which is one of the most common sources of problems -- one driver is not designed to handle a certain interleaving of function calls, while another is. So "it works on the developer's machine" -- maybe even on Nvidia and AMD -- but not on Intel.

Misha

19:04

@allquixotic Idk, I write OpenGL programs for $$$ and I haven't had these problems. The problems that I have had were performance problems with my geometry shaders.

sidran32

The API exposed is exactly the same, but the compilers and resulting bytecode will result in different assembler. We're not dealing with a spec for the architecture itself. Just the pipeline.

allquixotic

@sidran32 Performance isn't the only problem. The implementation of the drivers themselves can have quirks that are not directly a consequence of the hardware. It can be a simple semantic issue. I've hit it many a time.

For example the Mesa drivers on Linux try to adhere to the letter of the specification as closely as possible, which is the main source of their rendering and crash issues, because so many programs either try to work around bugs in proprietary drivers, or intentionally have non-spec-compliant code.

sidran32

@allquixotic Sometimes drivers have to compensate for a bug in the hardware itself, and then have to make some concessions. In order to make a GPU that complies to the DX spec, for instance, you have to pass WHQL certification, which verifies that your chip will render a predefined scene within some tolerance to the spec that Microsoft specifies.

allquixotic

Spec interpretation is a HUGE problem in driver implementation.

sidran32

Both ATI, NVidia, and Intel have to be able to run these tests and render these images as the spec requires.

Microsoft provides a software renderer that's used as a reference renderer, that's designed to be exact to the spec (again, within tolerance, as tolerance is also defined in the spec for certain things).

If you deviate, you don't get WHQL certification, which is important.

allquixotic

19:06

@sidran32 But the tests aren't exhaustive, and there are still a great number of quirks both in drivers and programs, where one tries to work around the other, or sometimes it even gets so tangled that a program tries to work around a driver, then the driver has to work around the workaround, then the program has to re-work around the driver's workaround, etc.

r.tanner.f

where did that come from? o_o

allquixotic

You're trying to tell me no WHQL-certified graphics drivers ever have any bugs, which is nonsense. I can show you a screenshot of one. It was a question I asked on SU, in fact.

Shiki

lol. ATI's website is giving me virus reports in Avast. Cool. I hate ATI.

sidran32

@allquixotic The tests are very numerous, but yes, not exhaustive. But, it covers most things. Like I said, the differences are more because of differences in architecture but also other corner cases.

allquixotic

2

Q: Why Does Text Get Squished Up In Word 2010 With A Translucent Watermark?

You may experience an anomaly in Word 2010 that looks like this: The expected result is supposed to be normally-displayed text. In the picture below, the text that is circled contains approximately 35 characters when rendered correctly. It is squished together to the point of being illegible in ...

That is a WHQL-certified (production/enterprise quality, whatever you wanna call it) driver bug.

sidran32

19:07

@allquixotic No, they can be WHQL qualified but still have bugs.

allquixotic

What I'm trying to say is, no finite amount of empirical testing for correctness can exhaustively determine that two people (or two different organizations) are interpreting the English-language specification of an API in precisely the same way.

sidran32

Sure, there are quirks but it's not really due to the vagueness of the spec. I've read the DX spec (for 9, 10 and 11, anyway) and had to use them as I worked on this at AMD, for the texturing subsystem of their GPUs. The spec is very precise. If there is ambiguity, we would shoot Microsoft a message and hammer it out.

allquixotic

One problem is empirical and one problem is conceptual. They can't interact like that.

Shiki

btw. that bug keeps both laptops Win8 free. Not like I would mind as it's a downgrade, not an upgrade.

allquixotic

Specifications are like laws, in that, being specified in English, depend on the way that the reader interprets the semantics (definitions) of the words and phrases contained within the spec.

sidran32

19:09

If there's room for variance, they'd provide a tolerance for considering the implementation to the spec.

And then we'd run against the WHQL tests, and they all have to pass.

(alongside our own tests, of course)

allquixotic

The only way to have a truly rigorous spec would be to specify it using something like a context-free grammar, kinda like a programming language, but for a spec.

sidran32

@allquixotic Have you read the DX specs? It's basically a white paper with tons of math equations, explaining how things should work and what should result when this or that happens.

allquixotic

You would basically have to specify the discrete finite automata that determines the possible bitmap raster states emitted by the driver (or in the case of GPGPU, the expected output for every possible input)

sidran32

Heh, they even have a warning section at the top saying "this is not light reading" :P

allquixotic

@sidran32 If it contains any English words and doesn't define them in terms of mathematics, it's still reliant on the English language. RFCs for example explicitly define "may", "should", "cannot", and so on -- even if they do that, though, there is still room for interpretation.

sidran32

19:11

It defines everything in terms of mathematics.

Quite literally. It's a very technical document.

allquixotic

Well, I haven't seen the DirectX spec, but the OpenGL specs are pretty heavily reliant on the English language.

sidran32

Like I said, if there was any ambiguity that caused question, we brought it up the chain to Microsoft and had them clarify. Sometimes there were back and forth arguments and they'd revise the spec based on our feedback.

allquixotic

Many a time, they end up posting a clarification or a minor revision to a specification for some extension, because someone asked a question where they didn't understand the meaning of one word that changed the entire meaning of the spec.

sidran32

I don't know about OGL, personally. DX, being the major player right now, got more attention. But OGL compliance wasn't too different from DX compliance.

There were maybe a couple different minor modes in the hardware and then the rest was handled in the driver.

allquixotic

The problem is, you also get tangled up with compatibility support: sometimes a driver contains a bug in an extension implementation and you want to continue to support that while also supporting the "right" way, and when you introduce a GPU family (say, from Intel) that doesn't normally get special treatment/testing by app engineers, you basically can't help but have an unstable program.

sidran32

19:14

With extensions, that's also more an OGL thing. Your concerns are probably valid, but I'd say more for OGL than DX :P

allquixotic

DirectX 9 has a ridiculous number of extensions.

Just like OpenGL 2.x.

sidran32

I know it did. Though I don't think DX programs rely on extensions quite as much as OGL programs tend to, simply because DX iterates more quickly of late.

allquixotic

DX10 and DX11 do move a lot of stuff into the "core", but OpenGL 3.x and 4.x did the same. The extensions (especially vendor-proprietary ones) are starting to slowly fade away on both fronts, mainly thanks to the loud complaints of app developers who use them and are then unpleasantly surprised when it doesn't work on all hardware.

sidran32

Yeah, they did. OGL is more or less on par with DX these days. But the big issue, historically, was that OGL stagnated with version 2.x for a long time, while DX continued to evolve and introduce new features. Hence the heavy reliance on ARB extensions.

allquixotic

Yeah.... and NV extensions, and ATI extensions...

sidran32

19:17

No, just ARB :P

opengl.org/archives/about/arb

allquixotic

Just wait until I show you the glxinfo for the Nvidia binary driver or even Mesa... the number of ATI_ and NV_ extensions is dizzying.

sidran32

Ok, maybe you're right :P Thought about that a bit and yeah, NV and ATI extensions probably didn't make it through the ARB either.

allquixotic

Lucky for you I'm not at my home computer with that info :P

sidran32

But yeah.

allquixotic

It takes up several pages

Also, OpenGL is essentially ClosedGL, because Floating Point Textures and S3TC are patented. I think they successfully worked around the S3TC patent because S2TC texture compression support is available in Ubuntu 12.10 by default even for U.S. users (S2TC is compatible with S3TC but uses a different algorithm), but the floating point issue remains

I've called it ClosedGL many times because it's patent-encumbered. You hate swpats, right? :P

sidran32

19:19

DX had those too but developers were slow to use them. I remember when ATI created tessellation. You could use it in DX9 with an extension, but rarely anyone did because it'd be more effort and it would mean making sure you don't try to use it when on NV hardware. And ATI weren't in bed with game developers like NV anyway, at the time, so they had little incentive.

Hehe, yeah, I know that texture compression was pretty much "this is owned by Microsoft, this is owned by NVidia, this is owned by ATI..." :P

allquixotic

Right now the big problem DX has (as far as I can tell) is trying to wrangle all the vendor-proprietary Stereosopic 3D implementations, although that's as much of the fault of the display manufacturers (Toshiba, Samsung, Panasonic, etc) as it is the driver authors.

As far as floating point textures, that patent is owned by HTC. I'm absolutely galled that they, being so reliant on open source software to use Android in their products, don't want to license that patent to OIN or otherwise give permission for the whole world to legally use floating point textures.

sidran32

I think that companies are consolidating around sequential frames/nvidia

allquixotic

<----- Never buying another HTC product

Floating point texture support is required to claim support for OpenGL 3.0, so the open drivers are stuck on OpenGL 2.x ("officially") until that's resolved.

Software patents, hindering, innovation. Welcome to 2012. :-/

sidran32

Yeah. I don't like software patents. Though it's hard when it comes to fixed-function GPU tech because it's technically not software, being baked into the GPU hardware itself. Though it's still math, and I'm not really down with that anyway. :P

allquixotic

It's not really fixed-function anymore, is it? Very little is still fixed function.

Remember that question/answer we were discussing the other day? :P

Drivers having to implement basically everything in terms of sending tasks to the SIMDs?

sidran32

19:24

@allquixotic There's still a ton of fixed-function in there. The balance is as you remove fixed functionality, you move more towards a CPU-like architecture, and lose performance gains.

Shaders aren't so much fixed function anymore. But that's only one part of the pipeline.

allquixotic

Oh well; it won't be long until the GPU is considered to basically be a SIMD coprocessor and has full access to coherent system memory cache through the CPU's MMU, and then you'll just have additional x86 instructions (or ARM, whatever) to tap into the massively parallel cores, and other instructions (the de facto) to tap into the cores better at serial tasks.

sidran32

Texture compression and filtering is still fixed-function (when done outside the shaders)

allquixotic

@sidran32 The patented side of texture compression is when you have an uncompressed texture, CPU-side, and need to compress it (using the CPU) with a certain algorithm, prior to uploading it to the GPU, to save on VRAM and PCI-e bus bandwidth. :/

That's what I beef with, not the fact that the hardware on the chip doing the decompression is patented. That's .... whatever.

sidran32

@allquixotic Yeah, that's driver work, and that's partially why it's hard to make an open-sourced GPU driver while protecting proprietary IP.

Which is a problem.

allquixotic

If they'd just make it so that the whole thing is a bulk fab hunk of SIMD cores, and drop the proprietary hardware blah, we could have fully open drivers (software patents notwithstanding), just like we have fully open CPUs.

I don't see how the actual electrical engineering design of GPUs makes them so unique that having them be as open as CPUs are today would be a real problem for the industry, aside from completing the transition to full-on GPGPU/HSA rather than fixed function.

sidran32

19:28

GPUs are crazy complex. :P

allquixotic

And anyway, once we reach that terminus, the rate of change in the GPU space should hopefully start to settle down.

sidran32

You can't just drop a bunch of SIMDs and accomplish what they can. Intel tried that and it didn't work out so well, IIRC.

And Sony, with Cell.

allquixotic

@sidran32 Larrabee? Yeah, I remember. Well, they based it on x86, which is a VM, rather than RISC; that was their first failure. :P

x86 needs to die. I'm sorry, I love it and all, but it needs to.

It's just soooo inefficient.

sidran32

Yeah, that's what I was thinking. Sony originally also wanted to use Cell for graphics, but they discovered it sucked, and had to quickly get a part from NVidia afterwards.

Hehe, yeah

allquixotic

I still think Larrabee, or something like it, is the eventual terminus of the evolution of microprocessor real-time 3d graphics (it'll all change again once quantum hits mainstream), and that you'll see a design similar to Larrabee stabilizing the GPU space much like we're stable on, basically, two processor arches right now (x86 and ARM).

From there you'll have incremental evolution, but nothing as ground-breaking as we see today.

OpenGL can't go on forever until OpenGL 89.2 starts implementing... I don't know, virtual reality shaders or some shit.

Ditto for D3D.

The backwards compatibility alone would take centuries to certify.

@sidran32 Looks like Intel is ramping up for an actual commercial release of the Larrabee design, as Intel Xeon Phi (branded), codename Knights Corner. en.wikipedia.org/wiki/Intel_MIC

Should be interesting to see if it can be used for real-time graphics or if it's just basically for throughput computing.

intel.com/content/www/us/en/high-performance-computing/…

It sounds rather GPU-like, but with enough CPU trappings to be programmable at a high level with a good multiprogramming framework, like OpenMP, or even a Java ThreadPoolExecutor with hundreds of threads.

allquixotic

20:56

our late answer review queue is huge :(

sidran32

21:15

@allquixotic Interesting. I thought they ditched it. Perhaps they only shelved it for a while?

allquixotic

@sidran32 They ditched the original Larrabee as far as the silicon and probably most of the design, but then a couple years later they made some improvements and are now well into the phase of productizing it. It'll be officially released as the Intel Xeon Phi in "no time at all" -- later this year or early next year.

sidran32

@allquixotic Ah, interesting

allquixotic

Once they announce a product name, that's it; it's going to production. So it's a reality, and they probably already have silicon of the final product.

sidran32

nods

allquixotic

I imagine it'll be extremely expensive, and I'm very interested to see how its performance per watt works out; but my gut feeling is that its design is for maximum parallelization of supercomputing workloads (high throughput, embarrassingly parallel workloads) optimizing for floor space, not power.

sidran32

21:18

Aha

We'll see

allquixotic

can't see it getting some kind of software Direct3D driver and being used as an off surface renderer for graphics (the off surface pixmaps could then be blitted to an Intel IGP / motherboard framebuffer)

it's just not designed for that, at least not efficiently

I still find it interesting though

it could be Intel's answer to Calxeda's many-ARM-CPU servers, just using their own IP rather than licensing ARM's. I had a feeling they wouldn't allow the density/power efficiency server market to fall out of their control without at least attempting to redeem themselves...

sidran32

nod

allquixotic

@sidran32 did you see earlier where, much to my chagrin, I satisfied the requirements of the windows-8 contest at the Level 1 tier? :/

sidran32

@allquixotic Nope. Congrats? :P

allquixotic

Congrats?! I'm a Linux cat, dude... it's almost embarrassing to satisfy the win-8 contest tiers.... :P

sidran32

21:24

@allquixotic Well then wear it ironically :P

allquixotic

I guess I let my penchant for answering juicy 0-answer questions get away from me and didn't check the tags.

it would be the height of irony if I got to level 4 in the next couple of days. hey, it's not impossible

sidran32

Level 4 is hard :P

I've been stagnant at 66% of the way to level 3 for days.

allquixotic

yeah...

I only have one windows 8 installation that actually works, and I probably don't have the right software or hardware configuration to even determine an answer to over half of the questions being asked

sidran32

Not that I'm trying, of course :P

allquixotic

my best bet is probably to ask some questions after playing around with, and breaking, Windows 8

sidran32

21:26

Heh

allquixotic

it's easier to break it than to figure it out :P

sidran32

Of course :P

Historically, I would break computers to figure them out. Usually what would happen as a child was I'd mess something up, and then since it's my parents' computer, I'd have to figure out how to fix it.

allquixotic

considering that my dad is now using Windows 8 as his full time production box, I think it's to my benefit to go ahead and learn a little bit about it, so at least I can quelch his belly aching... god that drives me up a wall

yeah, I did the exact same!

sidran32

For instance, that's how I figured out that Windows 3.1 stored screensaver passwords in a plaintext file that you could just edit in MSDOS :P

allquixotic

I got into programming because as a little kid my dad bought me a Lindows CD and made me install it (I was all into Windows at the time and didn't think it was worth my time, but he insisted), and I still wanted to play games so I ended up figuring out how to dual boot

but to dual boot I had to recompile LILO from source, so I started learning how to compile things

and then ended up making about 7000 posts on the Lindows/Linspire forums answering peoples' questions and many of the answers I had involved instructions for compiling stuff from source

sidran32

21:28

Cool

allquixotic

so I kind of graduated at some point from just recompiling stuff to actually making minor or not-so-minor edits to the source code to fix stuff

once you're down there building stuff, it's hard not to delve into the .c files (if you know a little C or Java) and fix whatever's broken, if it's easy

it all kind of came naturally

I was so excited to take AP Computer Science in 12th grade

sidran32

I always wanted to know how to program. I used to open up binary files in text editors and stare at them, curious what it all meant (before I learned anything).

Started by playing with DOS batch files.

Then in highschool I took actual programming courses

Matthew Wong

@allquixotic im taking ap comp sci too and im in 12th grade

allquixotic

yeah, I always wondered about how they make those DLL things that are so magical :)

@MatthewWong Cool! Back in my day, we learned C++ in AP Computer Science, and we used a Borland compiler! We were advised to bring in our program to compile on paper, so we could type it into the computer and wait for it to compile while the teacher lectured us :)

Even simple stdin/stdout programs took many minutes to compile

I bet your AP Comp Sci is based on Java now, eh?

I first learned how to implement linked lists by writing code on a blackboard with chalk...

using pointers in C++ and a simple struct

god, I feel old :/

Matthew Wong

@allquixotic yes it is :)

sidran32

21:37

@allquixotic I remember they went from Java to C++. I guess they went back to Java again?

allquixotic

well, I did my civic duty for the day; I voted until I had no more votes, and reviewed a bunch of questions, and only got 2 flag declines and a bunch of accepts :P

Matthew Wong

superuser.com/users/-1/community has a completed tile

sidran32

Before Java, I guess it was C++? And before that, I think it was FORTRAN, right?

allquixotic

@sidran32 they went from Java to C++?!?! O_o

sidran32

Yeah.

Matthew Wong

21:37

if it gets chosen does that mean none of us gets a chance in that tile?

sidran32

I had AP CS and it was supposedly the last year they were doing Java...

allquixotic

@MatthewWong bahahahaha... I imagine they are excluding user -1 from the drawing ;-)

@sidran32 wtf. I had AP CS and it was supposedly the last year they were doing C++!

sidran32

But my AP CS course was my 2003-2004 year in highschool

allquixotic

how does that work? we're the same age....

sidran32

Well, it was C++, then to Java, then back to C++

allquixotic

21:39

we should have graduated high school at the same time

OH

sidran32

AFAIK

allquixotic

I must have been the last year of the first round of C++ then

and the Java must've been very short-lived

sidran32

I guess

allquixotic

my teacher was like "next year we're doing Java"

sidran32

Of course, it could have been that it was just my college going from Java for all CS 101 courses into C++, irrespective of what the AP classes do.

I don't recall, now...

allquixotic

21:40

yeah, I'm not sure; but I thought the AP test was standardized?!

...to a single language and all!

sidran32

It is, but colleges have their own say on their own course structure

allquixotic

ah...

sidran32

I'm starting to think that maybe they just changed the CS 101 courses to C++ and AP didn't change.

allquixotic

the intro CS courses at Univ. of Maryland are/were also in Java

although I think they might switch to Python because it's "easier" and all the cool cats, CalTech and MIT are doing it :/

Python makes me want to physically damage the disk sector the interpreter binary is residing on :/

it's not only deleted, it's dead

sidran32

RIT did Java for the intro courses because it was supposed to be easy. But then they realized that it just made it that much harder for us to learn C++ in an abbreviated fashion later on in our progression to our degrees, and everything was mostly C++ after the 101 courses anyway, so they switched over to C++ again.

allquixotic

21:43

ah...

sidran32

It was harder, too. I had one C++ course, got a C, and still didn't figure out some of the core concepts until I took an optional 2 credit course called "C for C++ Programmers". After that, stuff clicked.

allquixotic

we had one course (Programming Languages -- what else?) where C++ was used. We used quite a lot of C in the low level courses (Computer Organization and Computer Architecture), but never C++ beyond Prog Langs.

for all the theoretical/mathy courses like algorithms we used Java.... for concurrency course, networking course, databases course, etc. we used Java

for operating systems... MIPS assembler trolololoool

well I say "concurrency course" but basically it was really advanced Java

everything from Java 5/6 generics, Java 6 locks, skip lists, lock-free and/or wait-free data structures, CMPXCHG (conceptual understanding of how it works) and atomic operations, various synchronization primitives both historical and current such as semaphores, memory barriers, mutexes, futexes, atomic integers, thread pools, etc

that was the most fun course for me : )

sidran32

Intro to CS 1-3 was all Java. Intro to CS 4 was basically "learn C++". After that, there was a Software Engineering required course that used Java, and Programming Language Concepts, that basically introduced you to other languages like lisp and other random ones, alongside "what makes a programming language". Then after that, anything else is all C++.

Networking, OS, System, Graphics, etc. All C++ programming.

Oh, except for Intro to Computer Architecture. We did MIPS ASM in that course :P

allquixotic

heh... CS 1-2 were "learn Java", programming languages was "learn Ruby, OCaml and C++", Programming Paradigms and Technologies (that "concurrency course") was "learn even more of the J2SE APIs", and the rest were math and theory :/

MIPS assembly for the win :/ well, not really

sidran32

Heh

allquixotic

21:48

I also found a truly awesome tie-in between a 400-level Philosophy of Logic and the 200-level CMSC Discrete Structures course

sidran32

My Encryption course had lots of programming too, but the professor didn't care what language you used. Just that you got the results and could show your code.

allquixotic

I was taking them concurrently and learning "boolean logic" one way in CMSC and in an entirely different way (predicate logic) in philosophy

sidran32

We had Discrete Math I and II, which taught mathematical logic, and was required for my CS degree.

allquixotic

then I learned it yet a third way with set theory

sidran32

Implications, sets, truth tables, formal proofs, etc.

allquixotic

21:49

my CMSC course was immediately before the philosophy course, too -- it was PERFECT

I learned the practical, hands-on, this-is-how-you-reason-in-CS stuff in computer science, then immediately went cross-campus to learn the philosophical underpinnings

they had all kinds of philosophical justifications and self-proofs of all the stuff underlying implication, sets, truth tables, etc

sidran32

That's cool, though that sounds a bit roundabout compared to how my school did it :P

allquixotic

roundabout maybe, but it gave me a fantastic mental model of the hierarchy of knowledge.. basically all scientific knowledge but especially digital logic and stuff

sidran32

It is cool

KronoS

anybody know asp.net?

allquixotic

in my assignments I would regularly make cross references to the other course and the professors would be like "that's really cool"

it was totally unintentional the way it worked out, too XD

pure coincidence

Matthew Wong

21:52

what does the number under our name represent? rep?

allquixotic

urgh, you've hit my achilles' heel, @KronoS... ASP.NET is the one semi-popular environment I really have no clue about. I know that it generally does the same types of things that JSP does (I know a good bit of JSP), but I don't know ASP or ASP.NET.

@MatthewWong yeah

it's your total SE-wide rep

btw, you are one lucky kid :3 I would love to be 17 learning AP Computer Science and have an amazing resource like StackExchange/StackOverflow/SuperUser to learn from

my "resources" were static HTML 1.0 webpages written by professors which basically read like prose... that, and textbooks

17..... that was.... NINE years ago o.o holy crap

almost 10 years

KronoS

@allquixotic you're getting old :P

Matthew Wong

well i can say that one of things im good at requires luck. (not coding)

allquixotic

:(((((

KronoS

(j/k) I'm right behind ya

sidran32

21:55

Yeah. Though when I was in highschool, I had QBasic in 9th grade, and after that, VB, Java, and C++ courses. However, due to what was available to me at home, I did all my hobbyist stuff in QBasic, and used a lot of guides and example programs from QBasic.com to learn stuff.

allquixotic

QBasic ahahaha YES... GOTO for the win

sidran32

I would code up cool tools and, as I was most familiar and comfortable with Java, otherwise, would try and reimplement some Java APIs in QBasic. Of course, with some approximation. :P

allquixotic

every once in a blue moon (well maybe less often than that even) when I'm programming I get a hankering for a good old GOTO.

Matthew Wong

i wish there was goto for php and java

allquixotic

@MatthewWong There is. Exception handling basically works exactly like goto or can be made to work as such

sidran32

21:56

I actually ended up making a library that, with simple function calls, could define a mouse-controlled GUI with everything except windows, and handled with a simple "ActionListener()" function call in a main loop.

allquixotic

You can skip over stack frames, ignore finalizers, and all kinds of ugly stuff with exceptions

also, finally

sidran32

I also tried making window managers in QBasic, but only with some success. Redrawing things behind windows as you moved them around was difficult to figure out.

I think Qbasic.com still exists...

allquixotic

@sidran32 awesome :D

yeah, you basically wrote your own rudimentary GUI toolkit in QBasic... that's praiseworthy

GTK isn't that far off from that ;) client-side windows are a little interesting on X11, but other than that, it's still a single threaded main event loop....

sidran32

But yeah. I ran up into the memory limit on QBasic quite a bit. Best thing I did was a Tetris clone that was mostly complete (except for some collision handling when rotating blocks). Had everything... high score list, progressing levels, cheats/tweaks, options that you can save, etc.

Yup

I don't know if I still have it. I broke it when I tried moving from storing the database info in strings to storing them in QBasic records, and never fixed it, though.

allquixotic

the thing that always got under my skin about event loops in general is that different toolkits and libraries have different ways of implementing them, and it's not always possible (or particularly easy) to integrate code from one library using one mainloop into another library using a different mainloop

sidran32

21:59

Wish I still had it.

allquixotic

you have to do something like add a callback into the mainloop to execute during every iteration and run the event pumping routine of the foreign mainloop

which is kind of manual and inelegant but works

works for Qt and Gtk together in-process...

sidran32

Oh, the fun thing about how I did the environment is that you could create the GUI with function calls, but you could also basically "serialize" it and store/load it from a file instead, skipping all the function call setups.

allquixotic

cool ... you also designed Glade 2! :O

sidran32

Haha

What's Glade 2? :P

allquixotic

libglade2, IIRC, used a binary representation of the serialized GUI once it was instantiated in memory

whereas new glade uses XML

sidran32

22:00

Aha neat

I don't even remember if it was a binary file or plaintext. But yeah. :P

This was way back in highschool. Junior or senior year.

allquixotic

no no no... glade version 1 did the binary thing; glade version 2 did XML; glade 3 improved on the XML; and GTK itself integrated Glade's basic functionality as a new class, GtkBuilder, more recently.

sidran32

Aha

allquixotic

anyway it's cool, I never polished off my projects as much as it sounds you have

I'm too much of a dabbler.... I'll jump into something with tremendous enthusiasm, spend a day or three on it, get it to where it just kind of does the basic proof of concept of what I want, and move on

Matthew Wong

so close to completing level 4 :(

allquixotic

except for a few things like rbpitch where I try to actually make it so that other people can use it...

sidran32

22:03

@allquixotic I was a dabbler too. I had a bunch of unfinished projects. But a couple, like that one, a pong clone, and nearly the tetris clone, were more or less complete.

The sysadmin at school actually let me load the Tetris program onto the network so other people could play it. I had it launch with a dos batch file and made it so when the program quit, it exited the interpreter, so you never had to look at the code itself.

allquixotic

the most complex project I can recall doing at a young age was while I was interning at Honeywell in 12th grade... I wrote and deployed a Microsoft Access application backed by SQL Server (with IT's permission) for managing a version-controlled document approval system for this certain type of administrative document that needed signatures...

had it insert an image of their signature (scanned) on the dotted line, re-created the paper form faithfully with interactive data controls, stored it all in SQL Server and managed "revisions" and re-approvals

all coded to a fairly loose spec given by non-technical people, so I did requirements capture all in my head and my scrawled notes, and implemented it all and shipped it

sidran32

I really need to go digging around and see if I have these programs sitting on floppies somewhere

allquixotic

about 12,000 lines of VBA in Microsoft Access, lots of ADO

sidran32

Cool

Wow

KronoS

@allquixotic wow intense

sidran32

22:05

Impressive for 12th grade :P

allquixotic

of course I didn't know how to actually go through a proper software engineering procedure, but I was like 18 years old, and practiced "persistence", "grinding through it", "asking on ExpertsExchange", and "overtime"

probably could've cut the code complexity by a lot if I really knew what I was doing, but from what I heard the next summer, they were still using it

pretty sure that, by now, IT has "rolled out" some big enterprise ERP system that completely encompasses all needs that every office bee will ever have, ever, and they now use that (probably...) for the thing my system did

BUT, it filled a role, at a point in time. :3

but that experience has given me a pretty much indelible stroke of genius when answering SU questions involving VBA... I rarely if ever use it for my own stuff anymore, but questions DO get asked about it, and I can pretty much rattle off the answer as 99% working "air code"

sidran32

Neat

I gotta go home. So, see ya!

allquixotic

I'm going home too!

laters!

Bob

@HackToHell IIRC Win7 would not install the 32-bit version to a GPT drive. I dunno if that's right or if Win8 is the same.

@HackToHell If you're on Windows, you're supposed to mark the partition as active (on MBR drives). diskpart and Disk Management can do that.

@Hennes in theory*

stupid fakeRAID actually reads slower for me

apart from some hardware redundancy, the main point was faster reads. Well, that failed.

Blog bot

22:21

Blog - Stack Exchange: Stack Exchange partners with Denver Startup Week

Bethany Marzewski on November 02, 2012

The Mile High City was buzzing last week as it hosted its first ever Denver Startup Week. With more than 80 events hosted by startups all around the city, the week attracted hundreds of entrepreneurs from the Colorado area. Since our Denver office just opened in August, this was a perfect time for us to get to know some of our new neighbors a little better.

Startup Job Fair

On Tuesday, October 23, after our quick spot on Channel 9 News, we set up shop at the Startup Job Fair to recruit even more awesome talent to our company. After two hours, we brought in nearly 50 resumes—thanks to everyone wh …

Bob

@allquixotic I'd argue that it is safer - you might get a couple of corrupted bits, which translates to a few corrupted files/FS metadata/folders. You can probably recover most of it. With RAID 0 failure, you're losing a good 1/nth of the array permanently, with no hope of recovery.