I made some 'debugging' macros to help in some situations for my code, but I don't know how to display line number (no error should be signaled, just message..). In NASM (like C) this is possible via __LINE__, but I would like to tell the line where the macro is called, because having the line in the macro is pointless. Is there a way to do this in Fasm?

Quiz2: I know I asked this before but wouldn't be nice to have Fasm a err directive, just like display, but displays, for example:


displays
Using

displays the message, but it doesn't make it a 'error' message and plus it also displays "Error: illegal instruction". So, why not add such a 'feature', it might look a bit more 'professional', or somesuch?

PS: Error should be signaled at assembly stage (maybe '@err' for preprocessor??)

http://board.flatassembler.net/topic.php?t=1154
As for the errors, the reason why not add it were already explained in that other thread where you asked about it.

Thanks for the guide.
EDIT: Why not make this official in next Fasm version (%L, etc.)?

about 'err', well, would it be hard to recognize 'err' as a 'valid' instruction that just prints the message as an erroneous message? Instead of telling 'Error: illegal instruction', just display the message? sorry if I'm lost in this area.. i didn't quite understand the other reply.

EDIT2: how to add %L (or other symbol) display the line that invoked the macro? this is what i really need, in fact. thanks

why not add these in fasm? any reason?

there is "display" for that.

If you meant that "err" should also end assembling at same time, it's imposible because of multi-pass assembling, code like this would cause error 'a' anyway in first pass, when it doesn't know that symbol "x" is used.

right now, 'err' stops assembly and says "Error: illegal instruction", or something like that.

but, instead of coding:



why not do

I know, it can be done via macros, but it will display 'a' as a message, not an error.. and it says 'illegal instruction' because it found 'err'.. i said, wouldn't it be nice to have 'err' do the same as an 'illegal instruction', but replace that message with the specified message? like err 'a' should display the error "Error: a"? okay, this is a minor feature, but it would be nice to have it..

About %L, or whatever, is there any reason why not add it?
thanks vid, you cleared that multi-pass thing a bit now

Whoops, i made mistake i meant "if ~ used x". Condition "if ~used x" is false, because symbol x is used later in source, but in first pass it will be predicted as true, so in this case err will stop on code that shouldn't be assembled.

yeah, I realized that.. that's why I used '~ used'
i get a lot of these errors, too. it happens all the time

EDIT: how does 'used' exactly work? I know only 'defined' that checks if variable is accessible from current position, but 'used'? thx

Same as "defined'. During parsing all symbols (non-reserved words) in source are collected into table. Then during assembling if assembler sees any definition of symbol (a:, label a, load a, ... ) then it marks that symbol has been used in this pass. If it see it used (anything but definition) then it marks it has been used.

A samples of various error signalizings:

this signalizes the error the first time it meets it during assembly - error happens.


this signalizes the error only when even after resolving the rest of code this one fails - thus you'll get the error only when you remove the "x:".


this causes the error during preprocessing.

Adding at least three new directives would cause even more confusion for the newcomers. I try now to write some new article for documentation about the fasm's programming layers, to cover many of such problems and questions.

Tomasz: Do you plan to add %L symbol? or %ML (which is the line that 'invoked' the macro).. maybe in next release? it could be very useful.. any reason for not adding it?


I don't think it would be that confusing, because these directives can be placed in another section of manual, and newcomers don't need these directives anyway (and these error directives are not really part of standard Fasm syntax, should be pretty easy to read, since they are not used much)... maybe when newcomers get more 'used' to Fasm and start doing some neat macroinstructions

With my personal experience with newcomers I'd rather be very careful about it - especially when feature seems to be so easy to use.

As for the %L, I was never convinced it would be actually so useful, when you don't from which file this line came. And file name cannot be handled this way at assembly time, since assembler operates only on numerical values assigned to symbols.

why not add preprocessor default 'symbols' then? like __LINE__ I suggested, and __FILE__. this helps me alot with some 'debug' macros i wrote.

Because adding such new class symbols would make another feature interacting with existing ones in a different way, an I'm already tired enough of explaining such interactions in the current already over-grown complexity of various layers in fasm. In initial design such complexity was not planned at all (since I didn't know fasm is going to be so widely used and universal) and therefore this design a bit limits what can be easily done, and what not. As only true solution I was thinking about writing a new assembler from scratch that wouldn't have such limitations, but I'm afraid it would be very much different assembler, and much slower than fasm is.

if you don't plan on adding __FILE__, it can be done manually in every source:

but %L or %ML are really needed, it cannot be done easily manually. pls add them

btw: didn't know it can be that complex to add some simple symbolic constants automatically (like __FILE__)

I just don't like it enough to force myself to implement it. At least at this moment.

i don't know how i can make my macros display that info and where it occured without %L.. i guess i'll have to figure out anyhow then

Sorry for that tone, I'm just a bit tired today, after many hours of non assembly-related work now trying to put at least few new lines into planned additions to fasm's documentation...

just read the new documentation yesterday, and it provided a clear explanation of numerical constants, it will definetely help newcomers (most of them look at examples first anyway I did).
keep up the good work

Yeah, we all feel that way sometimes, perfectly understandable.