Assembly programming and Windows

[TmX]

After visiting several assembly programming forums, I got an impression that most of the assembly programmers are using Windows.
Indeed there are lots of assembler available for Windows: MASM32, GoAsm, FASM, NASM, YASM, POASM, TASM, LZASM, etc.

While on non-Windows OSes (Linux, BSD, whatever), you're probably expected to use "C" (and a few assembly).

Is this just because Windows is still the most popular OS (and therefor lots of different assemblers are available) ?
Any other reasons?

::)

[dedndave]

unix-based OS's are written in C, for C programmers
a few of them write assembler also - but they have what they need
(i think the Gnu Assembler, aka "Gas", is somehwat common)
also, their C compilers allow inline assembler - that is enough to make them happy
but, you are right - windows is on the majority of computers - more demand for a variety
i dunno about the mac - mac's used to have motorola 68K based CPU's (not sure if that is still true)

[BlackVortex]

Haha, dedndave, you information is a "little" outdated.  :green2

Nowadays Macs use Intel (for many years they used PowerPC)

EDIT: yeah, I hate Macs with a passion too

[dedndave]

that doesn't surprise me at all - lol
if i start to read something and see the word "mac" i usually close the window   :snooty:

EDIT - kind of a shame, though - the 68K was easy code

[ToutEnMasm]

Is this just because Windows is still the most popular OS (and therefor lots of different assemblers are available) ?
Any other reasons?

Is there another system ? .At the last news,personnal computers selled without windows was numbered ...none.

[PBrennick]

Dave,

EDIT - kind of a shame, though - the 68K was easy code

The 6809e was even better. I still miss it. Imagine a processor with tons of undocumented opcodes (because the specs were stolen, they say) and some that would allow direct interaction of different size registers.

Paul

[mitchi]

Portable code is highly regarded in the GNU/Linux culture. C is very portable so it's really the language of choice on these platforms.
I know that GNOME is entirely in C while KDE is QT C++.

If I was to program for Linux, I would just use agner fog's objconv and a good assembler like sol_asm or fasm.

[dedndave]

yes Paul - i have written some code for the 6805 microcontrollers - very easy and fun
i think i prefer it over 8051 or PICs (do they even make PICs anymore ? - lol)

it is good to see we have a wide variety of favorite assemblers right here in our group
once i have learned my way around the API a little more, i might give GoAsm a try
although, Fasm and Nasm looked easy to use also (easy is not really what i look for)
there are so many toys out there - a lot to choose from

[raymond]

The Z-80 was also fun to program. It only had about 800 opcodes (quite easy to write your own assembler). The feature I liked the most was that it had two sets of data registers and you could switch at will between the two sets without loosing the data in them.

[dedndave]

yes - programming those liitle ones was an interesting challenge - especially if you had to do some math
math on those chips will quickly seperate the real coders from the script-kiddies - no fpu, here, folks

[Neo]

unix-based OS's are written in C, for C programmers
a few of them write assembler also - but they have what they need
(i think the Gnu Assembler, aka "Gas", is somehwat common)
also, their C compilers allow inline assembler - that is enough to make them happy
but, you are right - windows is on the majority of computers - more demand for a variety
i dunno about the mac - mac's used to have motorola 68K based CPU's (not sure if that is still true)

I've thought quite a bit about why there are so few unix/linux assembly programmers, and upon looking into it, it seems almost like linux and the Gnu Assembler are intended as a punch to the gut of anyone thinking of writing assembly for them.

<rant>The linux kernel has the problem that it tries to avoid assembly at all costs, resulting in massively-complicated chunks of C code to try to avoid just a couple of lines of assembly, and honestly, with the horrible non-standard syntax that Gas uses (made by AT&T, not any company that makes x86 chips), I don't blame them.  For examples of this paranoid avoidance of assembly, just look at anything to do with thread scheduling, like sched.c.  You can write a full thread scheduler in under 400 lines of assembly; why do they use almost 10,000 lines of C for just the thread scheduler abstraction layer?

Then, there's the added problem of system calls still being implemented with int 80h; it's a throwback to the days of DOS, where you had to put values into registers to indicate what functions you actually wanted to call instead of just calling a function.  They could at least have something equivalent to the insertion of jump tables (instead inserting code that puts the constant into a register and does int 80h), but no, they want to make assembly programming as difficult as possible.  Apparently linux now supports doing system calls with the syscall/sysret or sysenter/sysexit instructions, but it's still the same issue of hardcoding register values, plus who knows if they even let applications use them directly?</rant>

*sigh* oh well  :'(

[MichaelW]

The linux kernel has the problem that it tries to avoid assembly at all costs, resulting in massively-complicated chunks of C code to try to avoid just a couple of lines of assembly

I see a lot of this, complex C code that compared to equivalent assembly code is harder to understand, harder to code, and more likely to contain errors. But it is portable, and it impresses other like-minded C programmers.

I agree that AT&T syntax is horrible, but GAS has supported Intel syntax for years:

http://sourceware.org/binutils/docs-2.19/as/i386_002dSyntax.html#i386_002dSyntax

There is nothing resembling a high-level syntax and compared to MASM the macro system is very weak, but all of the essentials are there, even for creating applications completely in assembly.

[Neil]

Wow! Paul, 6809e processor, that brings back memories. What a fantastic processor, it was used in a 'Dragon' computer in this country (UK) & in Tandy computers in the USA. I 've looked on my bookshelf & I've still got '6809 Assembly Language Programming' by Lance Leventhal, but as you say it was the undocumented stuff that was such great fun.

[Neo]

I agree that AT&T syntax is horrible, but GAS has supported Intel syntax for years

Hmm... That begs the question why on earth people still use AT&T syntax... Compatibility, I suppose.

[mitchi]

Good points Neo. I think just like you on this matter. Hehe, at least they haven't lost all their sanity.
Here, we can see Linus Torvalds defending the use of the goto (jmp) against a typical second-rate coder straight out of the schools.
http://kerneltrap.org/node/553/2131