PR: Fix syncing of line number and plugin list for breakpoints and add unit tests.

[csabella]

Fixes #2179

1. The unit tests were written before the change to make sure the change fixed the bug reported.
2. I refactored the code slightly to keep my change cleaner. I hope that's OK.
3. With the unit tests, in the clear_breakpoints method, I put a note regarding the del. It's not calling the __del__ in the class.

[ccordoba12]

Thanks @csabella for your contribution! As I asked you last time: is this an issue in our 3.x branch too? If so, please move your branch to 3.x.

@rlaverde, please review @csabella's work, given that you're also working with the debugger ;-)

[csabella]

@ccordoba12 Technically it is a bug in 3.x, but since it's from an older bug report, I didn't know if you wanted to treat it as an enhancement instead. I'll be more careful to use the right branch in the future.

[ccordoba12]

@csabella, if an issue is marked as a bug and is present in 3.x, please solve it in 3.x. Only if it's marked as an enhancement, it needs to be worked in master.

I'll be more careful to use the right branch in the future.

Don't worry about it. In any case, you can ask us in the corresponding issue if it needs to be solved in master or 3.x.

[csabella]

@ccordoba12 I had to make some changes to the tests for 3.x. I commented each line that I changed. Would it be possible to modify those lines before merging with master? The original version of the tests were passing under 4.0. Thanks!

[rlaverde]

LGTM

I understand the refactoring, It removes some unnecessary code

Thanks for all the tests

[rlaverde]

I had to make some changes to the tests for 3.x. I commented each line that I changed.

Maybe something like this, could do the trick:

from spyder import version_info

if version_info > (4,):
    ...

Although I'm not sure if it's the best way to do it

[rlaverde]

Btw, I'm not able to reproduce #2179, @csabella What was the error?

[ccordoba12]

Maybe something like this, could do the trick

I also think we should follow a similar path regarding different behaviors for Spyder 3 and master in the tests you added @csabella.

[csabella]

Btw, I'm not able to reproduce #2179, @csabella What was the error?

The error was that the red dot was being marked on a line indicating that it was a breakpoint, but the breakpoint wasn't listed in the breakpoint plugin.

To recreate:

1. Add some lines of code to an editor.
2. Double click or F12 a line to add a breakpoint. At this point, there's a red dot and it's in the list.
3. Double click again to remove the breakpoint. It's gone from both.
4. Now, on the same line, Shift-F12 or Shift-Double-Click for the edit condition dialog.
5. Cancel from this dialog.
6. The red dot remains, but the breakpoint isn't in the list on the plugin.

[csabella]

I can't figure out the Travis error. The only change I made between this commit and the previous commit was to add a check the Spyder version number on lines that had been commented out. I would think that would make them basically the same, but yet, the last commit passed the tests and this one doesn't.

[ccordoba12]

@csabella, I restarted the failing Travis slot. Sometimes we got segfaults in Travis that require a restart. But now that you're a core dev, you can restart the failing slots yourself :-)