I've noticed that line numbers are sometimes off, sometimes quite a bit when I receive a run-time error. I had one today, it specified the correct function that was responsible for generating the error, however the line number was way off, it reported line 1463 but the entire function in question goes from line 708 to line 785. There is nothing remotely related at line 1463, it's a completely unrelated procedure that could not have possibly been called at the point the runtime error was received. The list of procedure/function calls was accurate, just not the line number, which makes it very difficult to figure out exactly what part of the function had the problem.

I am using FPC 3.0.4rc1, compiling with the text IDE and I have option "Generate also backtrace line information" selected.

Does anyone know what might cause the line numbers to be incorrect?

James

James Richters wrote

Does anyone know what might cause the line numbers to be incorrect?

I don't know what causes it, but I can confirm I have seen this too on one of my systems.

Regards,
Graeme

2017-11-06 19:49 GMT-02:00 James Richters :

Didn't the unit have any include file?

-Flávio

I vaguely recall reading something to the sense of:
this could be related to the optimization level used;
I can't remember where I read that;

-l.

Didn't the unit have any include file?

My doesn't have any include files. It's just a collection of procedure and functions that I use in many different programs.. but the unit also uses a lot of other units, some are more of my own, others are standard FPC units.

I vaguely recall reading something to the sense of: this could be related to the optimization level used;

I have it set to optimization target processor and code generation target processor both set to 80386

I have some programs that produce a runtime error that is exact, but others are way off. I have not figured out what causes it to be anything but exact yet.

James

James Richters wrote

I've noticed that line numbers are sometimes off, sometimes quite a bit when I receive a run-time error. I had one today, it specified the correct function that was responsible for generating the error, however the line number was way off, it reported line 1463 but the entire function in question goes from line 708 to line 785. There is nothing remotely related at line 1463, it's a completely unrelated procedure that could not have possibly been called at the point the runtime error was received. The list of procedure/function calls was accurate, just not the line number, which makes it very difficult to figure out exactly what part of the function had the problem.

One situation where this happens, is when line-endings are inconsistent (e.g. on Mac OS X, CR instead of LF on some lines).

Regards,

Adriaan van Os