Expression expected when trying to access segments

[Odzer]

I did not find any previous threads on this in the forum.

I am loading the translated .asm file in the CPU Emulator but there is a "In line 14, Expression expected" error, which happens with both the BasicTest and the PointerTest. In BasicTest.asm, line 14 is @LCL, and in PointerTest it is @THAT.

It seems like it is not possible to access these, even though I also load the .tst script in the emulator. Does anyone know what may be the problem?

[ivant]

The problem is probably one or two lines above the reported one. if you still cannot find it, you can post the start of the asm file here so we can take a look.

[Odzer]

This is the beginning of the file, up to a few lines after line 14 where the error comes up:

@3030
D=A
@SP
M=M+1
A=M-1
M=D

@SP
M=M-1
A=M
D=M
@THIS
A=M+0
M=D

@3040
D=A
@SP
M=M+1
A=M-1
M=D

@SP
M=M-1
A=M
D=M
@THAT
A=M+1
M=D


Also the beginning of BasicTest.asm:

@10
D=A
@SP
M=M+1
A=M-1
M=D

@SP
M=M-1
A=M
D=M
@LCL
A=M+0
M=D

[ivant]

A=M+0 is wrong. There is no such instruction.

[Odzer]

This is just adding the index to M, which in this case is zero. But I removed it and there is still the same error.

[Odzer]

I am probably doing another unsupported operation with M somewhere. I will take a better look and see if it works after that, thanks for the help.

[ivant]

Even though it looks like expressions, these are actually names. A=M, A=M+1, M=M-1 are valid assembly command, while A=M+0, M=M-2, A = M (spaces are not allowed) are not. The book/course material specifies the valid ones.

The HACK assembly language is special in this sense. Most others would use commands like LDA, STA, BNZ, etc. The authors chose a more mnemonic approach, but it sometimes confuses people that it's a more complex language than it actually is.

[ivant]

Here is an excellent guide to help you with other pitfalls like that.

[Odzer]

Thanks, that will be useful for the remaining assignments.

[WBahn]

ivant wrote

Even though it looks like expressions, these are actually names. A=M, A=M+1, M=M-1 are valid assembly command, while A=M+0, M=M-2, A = M (spaces are not allowed) are not. The book/course material specifies the valid ones.

A = M is an allowed instruction since white space is ignored. Similarly A D = M - 1 is acceptable.

Also, the author's emulator will accept both D+M and M+D as the same mnemonic, although that is not strictly compliant with their own language definition.

Somewhat oddly, their emulator strictly enforces the order of the destination registers, meaning what while AM = D+1 is acceptable, MA = D+1 is not. This is particularly unfortunate because the order defined for the assembly language, AMD, is not the same as the order of the bits in the instruction word, ADM.

[ivant]

WBahn wrote

A = M is an allowed instruction since white space is ignored. Similarly A D = M - 1 is acceptable.

Thank you, I didn't know that. I don't think I like this feature. I think it obfuscates how assembly works even more.

[WBahn]

ivant wrote

WBahn wrote

A = M is an allowed instruction since white space is ignored. Similarly A D = M - 1 is acceptable.

Thank you, I didn't know that. I don't think I like this feature. I think it obfuscates how assembly works even more.

I agree. I don't mind allowing whitespace around the mnemonics, but AD and M-1 should be treated as what they are -- mnemonics.