Could you compile a program manually?

[Paul Conrad]

You could, but why?

"Any sort of work in VB6 is bound to provide several WTF moments." - Christian Graus

[c2423]

No reason you couldn't compile something by hand, but it wouldn't be a good use of your time. If you really must write binary code and see the end product you'd be much better off writing a C-like compiler and writing relatively high level code and compiling it automatically. Writing this stuff in C as opposed to machine code or binary would be orders of magnitude faster. Judging by a quick look at your profile, along with your previous similar question, it looks like you're trying to write your own OS of sorts - you might like to take a look at Minix[^] which is a simplified Unix OS - I had to choose part of it to rewrite for my degree, so it's not too hard to take a look at and work out how things like drivers work. As a side note, does anybody else ever feel like they're channelling the MS paperclip? "It looks like you're trying to write a letter", "It looks like you're trying to write an Operating System" Hope this helps, Chris

[Colin Mullikin]

John Simmons / outlaw programmer wrote:

Take this sage advice in the spirit in which it's given - don't be a dork.

:thumbsup: :laugh:

The United States invariably does the right thing, after having exhausted every other alternative. -Winston Churchill America is the only country that went from barbarism to decadence without civilization in between. -Oscar Wilde Wow, even the French showed a little more spine than that before they got their sh*t pushed in.[^] -Colin Mullikin

[Lost User]

Brandon T. H. wrote:

Is there some way to process a source-to-executable like a computer by hand?

Yes (well duh, computers can't compute anything that humans can't, they're just faster at it), and it's not even that hard (just time-consuming). Converting whatever-language source to assembly? Piece of cake. Assembling by hand? Conceptually simple, only requires looking things up in a table. Putting it in the right executable format? Ok that one is a little tricky to get right, but also not that hard.

Brandon T. H. wrote:

If so, could you point to some documents?

Sure. For x86+windows: Intel manuals[^] ModRM encoding (on sandpile.org)[^] PE 101[^]

Brandon T. H. wrote:

Could you also compile a device driver source to ".sys" or ".vxd"

I see no reason why this should be impossible.

[lewax00]

Here's a more generic solution: Everything a computer does is manipulating numbers through a handful of basic operations (i.e. arithmetic, like addition/subtraction, and logic, like and, or, etc.), and move them around in memory. If you can manage those things then you can do anything a computer can do, it will just be much slower if you do. From a more mathematical stand point, any modern computer can be modeled as a Turing Machine[^], and you should be able to do everything a Turing machine can (of course, you won't have access to an infinite tape but this task is not infinite so you should be fine there). So to sum it up: if your computer can, you can too.

[Roger Wright]

It can be done, but it will take some time. Here's a manual[^].

Will Rogers never met me.

[_beauw_]

When I look at some of the pie-in-the-sky news items about things like automatic parallelization, I get the sense that we are alarmingly close to the day when an animated character in an IDE says things like "it looks like you're trying to write an OS." I think this kind of thing is an example of what I call the "Hero Anti-Pattern," in which one supposedly brilliant developer abstracts everything (garbage collection, parallelization, "plumbing" code...) away from the rest of us. My gut tells me that self-proclaimed geniuses toiling away in a vacuum will probably never really keep the rest of us from having to worry about whatever it is they're trying to fix. Instead, what results is that these people get to shape the future of development, and rake in the concomitant glory and cash, while the rest of us just end up doing the same tasks (albeit in Hero-ized form).

[_beauw_]

I've done a lot of hand-assembly, i.e. translating assembly language into machine language (raw numbers). This ends up being very mechanical. There's not much thought involved. As far as hand-compilation of higher-level language code goes, that's somewhat mechanical, too, and most people with CS degrees will have done some of this as a part of their education. One of my articles[^] relies pretty heavily on hand-compiled C code, although I must also note that I artificially constrained my use of C to certain features, to make this chore easier.

[RedDk]

That's a potentially devastating thing to say to a four year old

[Lost User]

If you're interested in c++ you could download and build the clang compiler (or gcc if you're brave) then step through it as it compiles and links your program. Good luck.

[c2423]

The way I look at it, they need to be kept busy doing something while the rest of us make stuff that actually benefits people.

[Lost User]

He's not that new; and check out his signature. apologies if Brandon T.H is a girl.

One of these days I'm going to think of a really clever signature.

[Sentenryu]

remember me to get you a 5 when the vote system gets back online, you deserve it.

I'm brazilian and english (well, human languages in general) aren't my best skill, so, sorry by my english. (if you want we can speak in C# or VB.Net =p)

[loctrice]

Richard MacCutchan wrote:

hey presto, Robert's your mother's brother.

I punched the wife when I read that. I will never forget it.

If it moves, compile it

[loctrice]

:laugh:

If it moves, compile it

[John Atten]

Aren't we already almost there? VS: "File/New/ASP.NET MVC Project" or "File/New/Entity Model." Eclipse: "File/New/Java Package" Most modern IDE's abstract away so much boiler plate code that we have acheived almost the equivalent of Clippy already. Newcomers, such as myself, begin to believe that that's all there is to it. The "run" button, which in reality "builds" such projects then executes the build result, optionally stepping through in an integrated debugger, removes a whole lot behind-the-scenes operations. While all of these are great for productivity, and in many cases represent the ultimate goal of code re-use, they also deprive (for a time, anyway) us of the need to learn how these hidden parts of the process work.

[Isfeasachme]

Sage advice... Pssh. Ya junior code snot. I'm glad I'm not the only assembler dork out there. This is the sort of question asked by people who end up designing the compilers you all use.

[Member 110323]

Ah, the memories! First did this when my computer had a 16 bit address bus and 8 bit op-codes and a total of 4K of RAM. Then I discovered there was something called a "compiler" and haven't had the urge to engage in this kind of masochism since.

[Norm Powroz]

I used to do this all the time in my mainframe days. Operating systems used to be written in assembler, and the only way to correct bugs or add extensions on the fly was to hand-assemble the code into pure machine code. One system I supported for a number of years had an on-line patcher so you could patch the O/S while the system was up and running. I used to give the operations staff hairy fits when I did this ;) Just to show how big some leaps can be -- I went from an all-assembler O/S to a machine whose O/S was written in PL/1. An interesting leap indeed, especially since the source code of the PL/1 O/S was all on-line, so you could use parts of it in your own applications.

[gggustafson]

Many years ago, while teaching at a local college and working as an architect for a new database system, I decided to write the compiler I was requiring my students to write. At that time, I was considered a good assembler programmer. I decided to compare the output of the compiler I wrote with my own hand-coded assembly. As I recall, my assembly contained about 15 instructions; the compiler's generated assembly code contained 8 instructions. I learned a valuable lesson.

Gus Gustafson