Incorrect Debug Exception Highlighting with Multiple Files of the Same Name

[ghost]

Description

When an exception occurs, Overture attempts to highlight the line which the exception occurred on, however when a project contains multiple files with the same name in different directories, Overture displays the file in the first directory (alphabetically), and not the correct file.

Steps to Reproduce

1. Create a project with at least two directories, for instance, dir1/ and dir2/.
2. Create a .vdmsl file with the same name in each directory, for instance, dir1/type.vdmsl and dir2/type.vdmsl.
   
3. Create a valid file in dir1/type.vdmsl.
   
4. Create an invalid file / erroneous function in dir2/vdmsl.
   
5. Run the erroneous function.
   

Expected behavior: The debugger highlights the erroneous line in the file where the exception occurred.

Actual behavior: The debugger highlights the corresponding line in the first file it finds, matching the name of the erroneous file.

Reproduces how often: Always

Versions

Overture 2.7.2 Windows 10 x64

Additional Information

The issue was found when running the project as a console.

The exception chosen is one which occurs at runtime (Error 4: Cannot get bind values for type nat in 'DEFAULT'), although this is irrelavent for this bug.

[nickbattle]

Thanks for reporting this. I can confirm that it is reproducible on the latest Overture 2.7.4 (on Linux, but it is unlikely to be a platform issue). It is the same whether launched from the console or a launcher.

I'll take a look. But it looks like it's making a naive file lookup at some point.

If it's causing a problem, an obvious workaround is to rename the file, though clearly this should work as expected.

[nickbattle]

It is behaving correctly in the pure command line - that is, the stand alone Overture jar. This suggests the problem is with the IDE/VDM communication of the stack location.

Parsed 1 module in 0.185 secs. No syntax errors
Type checked 1 module in 0.187 secs. No type errors
Initialized 1 module in 0.408 secs. 
Interpreter started
> p f(1)
Error 4134: Infinite or NaN trouble in 'DEFAULT' (dir2/types.vdmsl) at line 6:14
Stopped in 'DEFAULT' (dir2/types.vdmsl) at line 6:14
6:  	f(a) == a/0;
[MainThread-3]> q
Exception: terminated
Runtime: Error 4134: Infinite or NaN trouble in 'DEFAULT' (dir2/types.vdmsl) at line 6:14
> 

[ghost]

I assumed this would be the case, apologies for not testing direct execution of the jar myself, I could have narrowed down the issue a little more for you.

I only came across the issue as I was revising for an exam on VDM, and ended up with 2 different files named set.vdmsl. This is where I received the initial runtime error, and was suprised when Overture shows the line in the wrong file.

I'd love to suggest a solution myself, but I'm completely unfamiliar with the eclipse framework, and only know the very basics of Java File IO.

[nickbattle]

I always test the stand-alone jar first, so I know whether it's the language engine or the IDE. If it's the language side, I'll know how to fix it :-)

It definitely looks like the IDE though, because the comms between the IDE and language engine includes the qualified file information:

>>	run -i 101
<<	<?xml version="1.0" encoding="UTF-8"?><response command="run" status="break" reason="error" transaction_id="101"><internal threadId="3" threadName="MainThread-3" threadState="RUNNING" /></response>
>>	stack_get -i 102
<<	<?xml version="1.0" encoding="UTF-8"?><response command="stack_get" transaction_id="102"><stack level="0" type="file" filename="file:/home/nick/FMIWorkspace/TestSL/dir2/types.vdmsl" lineno="6" cmdbegin="6:11"/></response>

So now I need to find someone who understands the IDE... :-(

[nickbattle]

Thanks Hugo!

[nickbattle]

I think there may be a problem with this fix.

I'm trying to do some "serious" debugging on the FMI spec, rather than the test example above. When it stops at a breakpoint and trying to open the source file concerned, it uses a bizarre paths like:

../../INTO-CPS-Association/FMI-VDM-Model/FMI3/ModelVariables_2.2.10.vdmsl line: 435	

I'll re-open this issue, as it looks like this is linked to the fix above?

[nickbattle]

(and those paths are not found, incidentally. So you get a blank editor pane with an invite to "Edit Source Path Lookup")

[nickbattle]

Easiest way to reproduce this is to load the FMI2 model (https://github.com/INTO-CPS-Association/FMI-VDM-Model/tree/master/FMI2), and put a breakpoint on the start of the tests() function at the end of Tests.vdmsl:

	tests: () +> seq of bool
	tests() ==
	[
		testCausalityVariabilityInitial(),      <---- add a BREAK on this line
		isValidFMIModelDescription(example),
		isValidFMIModelDescription(springMassDamper)
	]
	^	testStructuredNames()

	post elems RESULT = {true};

Then just run DEFAULT'tests(). You'll see it stop on the line, but the file is reported (in my case) as ../../INTO-CPS-Association/FMI-VDM-Model/FMI2/Tests/Tests.vdmsl. This file is not found.