Fix keyboard navigation commands

[Discookie]

Fixes#71.

[csordasmarton]

I have the following test files:
main.h:

#ifndefMAIN_H#defineMAIN_Hintfoo(intparam) {
returnparam;
}
#endif

main.cpp:

#include"main.h"intmain() {
return1 / foo(0);
}

If I analyze it CodeChecker will find the following report:

Let's assume that I click on the second bug step:

Now if I run the Next step command nothing happens:

I assumed that your patch fixes this use case so the editor will open main.h and select the correct ranges.

[Discookie]

That is a very good example for testing overlapping ranges, thank you!
Fixed it so it prioritizes exact matches over matching the step's range, but it still doesn't quite work in this case.

The issue is that the 2nd (calling foo) and 5th (returning from foo) steps have the exact same location. The extension has no memory of which step it's supposed to be on, so stepping backwards always steps from Calling and forwards from Returning.
Dealing with these exact overlaps is better fit for a separate issue I believe.

[csordasmarton]

I don't really see what kind of problem this PR solves. Can you please describe the steps what previously don't work but with your patch it will work with a code example?

Also if we don't want to fix the problem fully could you please create another issue for it?

[Discookie]

It solves a simpler version of the issue:

main.cpp

#include"main.h"intmain() {
returnfoo(0);
}

main.h

#ifndef MAIN_H
#defineMAIN_Hintfoo(int param) {
return1 / param;
}
#endif

In this case, navigation only worked inside main.h.
Also fixes range-checks in general, so the cursor can be anywhere in a report's range and the navigation will work.

Another added fix is for overlapping ranges:
( | is where the cursor jumps when you step to the repr. step.)

\-|----/ Step 1
|-/ Step 2
| Cursor, stepping backwards.

Before, stepping backwards from step 2 skipped range 1, and jumped to step 1's previous step. Now when the cursor is on step 2's exact position, that one will get prioritized.

Created #86 for the remaining case.

[csordasmarton]

LGTM!

[Discookie]

Fixes#71.

[csordasmarton]

I have the following test files:
main.h:

#ifndefMAIN_H#defineMAIN_Hintfoo(intparam) {
returnparam;
}
#endif

main.cpp:

#include"main.h"intmain() {
return1 / foo(0);
}

If I analyze it CodeChecker will find the following report:

Let's assume that I click on the second bug step:

Now if I run the Next step command nothing happens:

I assumed that your patch fixes this use case so the editor will open main.h and select the correct ranges.

[Discookie]

That is a very good example for testing overlapping ranges, thank you!
Fixed it so it prioritizes exact matches over matching the step's range, but it still doesn't quite work in this case.

The issue is that the 2nd (calling foo) and 5th (returning from foo) steps have the exact same location. The extension has no memory of which step it's supposed to be on, so stepping backwards always steps from Calling and forwards from Returning.
Dealing with these exact overlaps is better fit for a separate issue I believe.

[csordasmarton]

I don't really see what kind of problem this PR solves. Can you please describe the steps what previously don't work but with your patch it will work with a code example?

Also if we don't want to fix the problem fully could you please create another issue for it?

[Discookie]

It solves a simpler version of the issue:

main.cpp

#include"main.h"intmain() {
returnfoo(0);
}

main.h

#ifndef MAIN_H
#defineMAIN_Hintfoo(int param) {
return1 / param;
}
#endif

In this case, navigation only worked inside main.h.
Also fixes range-checks in general, so the cursor can be anywhere in a report's range and the navigation will work.

Another added fix is for overlapping ranges:
( | is where the cursor jumps when you step to the repr. step.)

\-|----/ Step 1
|-/ Step 2
| Cursor, stepping backwards.

Before, stepping backwards from step 2 skipped range 1, and jumped to step 1's previous step. Now when the cursor is on step 2's exact position, that one will get prioritized.

Created #86 for the remaining case.

[csordasmarton]

LGTM!

[Discookie]

Fixes#71.

[csordasmarton]

I have the following test files:
main.h:

#ifndefMAIN_H#defineMAIN_Hintfoo(intparam) {
returnparam;
}
#endif

main.cpp:

#include"main.h"intmain() {
return1 / foo(0);
}

If I analyze it CodeChecker will find the following report:

Let's assume that I click on the second bug step:

Now if I run the Next step command nothing happens:

I assumed that your patch fixes this use case so the editor will open main.h and select the correct ranges.

[Discookie]

That is a very good example for testing overlapping ranges, thank you!
Fixed it so it prioritizes exact matches over matching the step's range, but it still doesn't quite work in this case.

The issue is that the 2nd (calling foo) and 5th (returning from foo) steps have the exact same location. The extension has no memory of which step it's supposed to be on, so stepping backwards always steps from Calling and forwards from Returning.
Dealing with these exact overlaps is better fit for a separate issue I believe.

[csordasmarton]

I don't really see what kind of problem this PR solves. Can you please describe the steps what previously don't work but with your patch it will work with a code example?

Also if we don't want to fix the problem fully could you please create another issue for it?

[Discookie]

It solves a simpler version of the issue:

main.cpp

#include"main.h"intmain() {
returnfoo(0);
}

main.h

#ifndef MAIN_H
#defineMAIN_Hintfoo(int param) {
return1 / param;
}
#endif

In this case, navigation only worked inside main.h.
Also fixes range-checks in general, so the cursor can be anywhere in a report's range and the navigation will work.

Another added fix is for overlapping ranges:
( | is where the cursor jumps when you step to the repr. step.)

\-|----/ Step 1
|-/ Step 2
| Cursor, stepping backwards.

Before, stepping backwards from step 2 skipped range 1, and jumped to step 1's previous step. Now when the cursor is on step 2's exact position, that one will get prioritized.

Created #86 for the remaining case.

[csordasmarton]

LGTM!

[Discookie]

Fixes#71.

[csordasmarton]

I have the following test files:
main.h:

#ifndefMAIN_H#defineMAIN_Hintfoo(intparam) {
returnparam;
}
#endif

main.cpp:

#include"main.h"intmain() {
return1 / foo(0);
}

If I analyze it CodeChecker will find the following report:

Let's assume that I click on the second bug step:

Now if I run the Next step command nothing happens:

I assumed that your patch fixes this use case so the editor will open main.h and select the correct ranges.

[Discookie]

That is a very good example for testing overlapping ranges, thank you!
Fixed it so it prioritizes exact matches over matching the step's range, but it still doesn't quite work in this case.

The issue is that the 2nd (calling foo) and 5th (returning from foo) steps have the exact same location. The extension has no memory of which step it's supposed to be on, so stepping backwards always steps from Calling and forwards from Returning.
Dealing with these exact overlaps is better fit for a separate issue I believe.

[csordasmarton]

I don't really see what kind of problem this PR solves. Can you please describe the steps what previously don't work but with your patch it will work with a code example?

Also if we don't want to fix the problem fully could you please create another issue for it?

[Discookie]

It solves a simpler version of the issue:

main.cpp

#include"main.h"intmain() {
returnfoo(0);
}

main.h

#ifndef MAIN_H
#defineMAIN_Hintfoo(int param) {
return1 / param;
}
#endif

In this case, navigation only worked inside main.h.
Also fixes range-checks in general, so the cursor can be anywhere in a report's range and the navigation will work.

Another added fix is for overlapping ranges:
( | is where the cursor jumps when you step to the repr. step.)

\-|----/ Step 1
|-/ Step 2
| Cursor, stepping backwards.

Before, stepping backwards from step 2 skipped range 1, and jumped to step 1's previous step. Now when the cursor is on step 2's exact position, that one will get prioritized.

Created #86 for the remaining case.

[csordasmarton]

LGTM!

[Discookie]

Fixes#71.

[csordasmarton]

I have the following test files:
main.h:

#ifndefMAIN_H#defineMAIN_Hintfoo(intparam) {
returnparam;
}
#endif

main.cpp:

#include"main.h"intmain() {
return1 / foo(0);
}

If I analyze it CodeChecker will find the following report:

Let's assume that I click on the second bug step:

Now if I run the Next step command nothing happens:

I assumed that your patch fixes this use case so the editor will open main.h and select the correct ranges.

[Discookie]

That is a very good example for testing overlapping ranges, thank you!
Fixed it so it prioritizes exact matches over matching the step's range, but it still doesn't quite work in this case.

The issue is that the 2nd (calling foo) and 5th (returning from foo) steps have the exact same location. The extension has no memory of which step it's supposed to be on, so stepping backwards always steps from Calling and forwards from Returning.
Dealing with these exact overlaps is better fit for a separate issue I believe.

[csordasmarton]

I don't really see what kind of problem this PR solves. Can you please describe the steps what previously don't work but with your patch it will work with a code example?

Also if we don't want to fix the problem fully could you please create another issue for it?

[Discookie]

It solves a simpler version of the issue:

main.cpp

#include"main.h"intmain() {
returnfoo(0);
}

main.h

#ifndef MAIN_H
#defineMAIN_Hintfoo(int param) {
return1 / param;
}
#endif

In this case, navigation only worked inside main.h.
Also fixes range-checks in general, so the cursor can be anywhere in a report's range and the navigation will work.

Another added fix is for overlapping ranges:
( | is where the cursor jumps when you step to the repr. step.)

\-|----/ Step 1
|-/ Step 2
| Cursor, stepping backwards.

Before, stepping backwards from step 2 skipped range 1, and jumped to step 1's previous step. Now when the cursor is on step 2's exact position, that one will get prioritized.

Created #86 for the remaining case.

[csordasmarton]

LGTM!

[Discookie]

Fixes#71.

[csordasmarton]

I have the following test files:
main.h:

#ifndefMAIN_H#defineMAIN_Hintfoo(intparam) {
returnparam;
}
#endif

main.cpp:

#include"main.h"intmain() {
return1 / foo(0);
}

If I analyze it CodeChecker will find the following report:

Let's assume that I click on the second bug step:

Now if I run the Next step command nothing happens:

I assumed that your patch fixes this use case so the editor will open main.h and select the correct ranges.

[Discookie]

That is a very good example for testing overlapping ranges, thank you!
Fixed it so it prioritizes exact matches over matching the step's range, but it still doesn't quite work in this case.

The issue is that the 2nd (calling foo) and 5th (returning from foo) steps have the exact same location. The extension has no memory of which step it's supposed to be on, so stepping backwards always steps from Calling and forwards from Returning.
Dealing with these exact overlaps is better fit for a separate issue I believe.

[csordasmarton]

I don't really see what kind of problem this PR solves. Can you please describe the steps what previously don't work but with your patch it will work with a code example?

Also if we don't want to fix the problem fully could you please create another issue for it?

[Discookie]

It solves a simpler version of the issue:

main.cpp

#include"main.h"intmain() {
returnfoo(0);
}

main.h

#ifndef MAIN_H
#defineMAIN_Hintfoo(int param) {
return1 / param;
}
#endif

In this case, navigation only worked inside main.h.
Also fixes range-checks in general, so the cursor can be anywhere in a report's range and the navigation will work.

Another added fix is for overlapping ranges:
( | is where the cursor jumps when you step to the repr. step.)

\-|----/ Step 1
|-/ Step 2
| Cursor, stepping backwards.

Before, stepping backwards from step 2 skipped range 1, and jumped to step 1's previous step. Now when the cursor is on step 2's exact position, that one will get prioritized.

Created #86 for the remaining case.

[csordasmarton]

LGTM!

[Discookie]

Fixes#71.

[csordasmarton]

I have the following test files:
main.h:

#ifndefMAIN_H#defineMAIN_Hintfoo(intparam) {
returnparam;
}
#endif

main.cpp:

#include"main.h"intmain() {
return1 / foo(0);
}

If I analyze it CodeChecker will find the following report:

Let's assume that I click on the second bug step:

Now if I run the Next step command nothing happens:

I assumed that your patch fixes this use case so the editor will open main.h and select the correct ranges.

[Discookie]

That is a very good example for testing overlapping ranges, thank you!
Fixed it so it prioritizes exact matches over matching the step's range, but it still doesn't quite work in this case.

The issue is that the 2nd (calling foo) and 5th (returning from foo) steps have the exact same location. The extension has no memory of which step it's supposed to be on, so stepping backwards always steps from Calling and forwards from Returning.
Dealing with these exact overlaps is better fit for a separate issue I believe.

[csordasmarton]

I don't really see what kind of problem this PR solves. Can you please describe the steps what previously don't work but with your patch it will work with a code example?

Also if we don't want to fix the problem fully could you please create another issue for it?

[Discookie]

It solves a simpler version of the issue:

main.cpp

#include"main.h"intmain() {
returnfoo(0);
}

main.h

#ifndef MAIN_H
#defineMAIN_Hintfoo(int param) {
return1 / param;
}
#endif

In this case, navigation only worked inside main.h.
Also fixes range-checks in general, so the cursor can be anywhere in a report's range and the navigation will work.

Another added fix is for overlapping ranges:
( | is where the cursor jumps when you step to the repr. step.)

\-|----/ Step 1
|-/ Step 2
| Cursor, stepping backwards.

Before, stepping backwards from step 2 skipped range 1, and jumped to step 1's previous step. Now when the cursor is on step 2's exact position, that one will get prioritized.

Created #86 for the remaining case.

[csordasmarton]

LGTM!

[Discookie]

Fixes#71.

[csordasmarton]

I have the following test files:
main.h:

#ifndefMAIN_H#defineMAIN_Hintfoo(intparam) {
returnparam;
}
#endif

main.cpp:

#include"main.h"intmain() {
return1 / foo(0);
}

If I analyze it CodeChecker will find the following report:

Let's assume that I click on the second bug step:

Now if I run the Next step command nothing happens:

I assumed that your patch fixes this use case so the editor will open main.h and select the correct ranges.

[Discookie]

That is a very good example for testing overlapping ranges, thank you!
Fixed it so it prioritizes exact matches over matching the step's range, but it still doesn't quite work in this case.

The issue is that the 2nd (calling foo) and 5th (returning from foo) steps have the exact same location. The extension has no memory of which step it's supposed to be on, so stepping backwards always steps from Calling and forwards from Returning.
Dealing with these exact overlaps is better fit for a separate issue I believe.

[csordasmarton]

I don't really see what kind of problem this PR solves. Can you please describe the steps what previously don't work but with your patch it will work with a code example?

Also if we don't want to fix the problem fully could you please create another issue for it?

[Discookie]

It solves a simpler version of the issue:

main.cpp

#include"main.h"intmain() {
returnfoo(0);
}

main.h

#ifndef MAIN_H
#defineMAIN_Hintfoo(int param) {
return1 / param;
}
#endif

In this case, navigation only worked inside main.h.
Also fixes range-checks in general, so the cursor can be anywhere in a report's range and the navigation will work.

Another added fix is for overlapping ranges:
( | is where the cursor jumps when you step to the repr. step.)

\-|----/ Step 1
|-/ Step 2
| Cursor, stepping backwards.

Before, stepping backwards from step 2 skipped range 1, and jumped to step 1's previous step. Now when the cursor is on step 2's exact position, that one will get prioritized.

Created #86 for the remaining case.

[csordasmarton]

LGTM!