I wonder what ASM syntax do you think is better? Here's the way I see it:

Intel Syntax
Good:

• No cryptic, hard-to-type %s and $s everywhere.
• Intuitive operator order (dest, source)

Bad:

• You have to manually say the size of the data if the assembler can't figure it out

AT&T Syntax
Good:

• Operator suffixes to explicitly say the data size you want.

Bad:

• Cryptic-looking, difficult-to-type %s and $s everywhere
• Confusing operator order (source, dest)

Personally I prefer Intel.. but that's probably just because I started out doing 16 bit asm on Windows. It's just what you get used to..

I prefer gas syntax despite learning Intel syntax and writing a lot of code in Intel syntax long before learning gas syntax.

Switching between two operand orders is confusing. Neither order is inherently more or less confusing.

I'm not thrilled with the % or many other ways that gas syntax tries to be more machine independent than an assembler can actually be.

But typed and sized memory reference operands are just so wrong for the concept of assembler programming that they overwhelm everything else.

When the assembler can't deduce the size or type of an operand from the register choice or instruction or other local info, that info should be provided as a suffix to the opcode. Anything else is just wrong. The Intel method of defining type and size info for memory addresses is a bad way of pretending that asm is a high level language. If you want to program in C, program in C. If you want to program in asm, a name defined for an address should just represent an address, not an address type and size.

Maybe someone should make a new assembly syntax, one that is like intel with automatic operand sizing.

@johnsfine

Personally I find the destination-first syntax easier to understand, maybe because I'm used to C, and because in mov instructions, for example, I can at a glance see what the register/memory location is set to by looking at the end of the line.

Also I find typing those awkwardly-placed '%' and '$' characters so often very difficult, and with time maybe a bit painful. I also think they add a lot of "visual noise", severely reducing the code's readability.

And I find that [] is more intuitive than () for dereferencing for me, again because I am used to C. But I don't really care, this is more of a little thing.

But OTOH I really like AT&T Syntax's size suffixes. You're right that it is a little wrong for the assembler to remember the types and sizes of "variables".

@smeezekitty

Wouldn't it be great if there would be an assenbly language with instruction suffixes, destination-first operator order, and no symbols before registers/literals?

Assembly syntax is a non-issue nowadays. One typically writes in assembly only reset routine and ISRs which comprise a very small part of the code present in a system.

Complaining about assembly syntax doesn't mane sense in the presence of languages like Perl/Python/etc - one can write a layer of assembly text generating subs in them and choose for the subs the desired order of operands (src, dst, whatever else).

Or just plain parser that would accept the desired assembly syntax.

@sergei: some people write assembly for fun not because they have to

Of course; by the same token they can write for fun an assembly compiler or a preprocessor.

Exactly.

I might, but I don't know how.

You have no chance of knowing - because whenever you are supposed to make an effort and to read, comprehend and analyze, you skip.

Sergei;
Please do not post personal comments about other members in this way. Perhaps a better reply would be to ask if MTK wants to learn?

MTK;
Please consider that your reply to Sergei was inviting a retort.

Both of you----please look for opportunities to NOT bicker with each other.

Hmm. I think the AT&T order (source, dest) is more intuitive. I can see where your coming from, if you treat it like an assignment operator in higher-level languages, but I don't treat it that way.

Yes, I look at it like an assignment operator. I am also used to C's memory and string functions, that take the destination first.

I also find it easier to see at a glance what the register is being assigned to when the source is at the end of the line.

I agree.
But it would also be trivial to have automatic sizing.
Both syntax's are flawed and outdated.

Compare:

Intel:

AT&T:

My idea:

Or maybe "d" for "dword" makes more sense than "l" for "long"?