Fix issues with grouped tests

[dellaert]

Fixes #91 and hopefully addresses some issues in #1102

[dellaert]

Hey @mikesheffler, this works now. If you review I can merge, and then you can merge in develop to see the effect on Windows.

[varunagrawal]

All we had to do was add namespace guards to all the global variables in the tests??? Sheesh, I'm kinda sad that #91 took this long to resolve. LGTM!!!

[varunagrawal]

Any reason we've killed these two tests?

[varunagrawal]

Ah nvm my previous comment. I saw additional serialization tests being removed and then realized I should scroll down a bit.

[mikesheffler]

Hey @mikesheffler, this works now. If you review I can merge, and then you can merge in develop to see the effect on Windows.

Sorry, I've been tied up all day.

In plain English, what's the goal here? Are we trying to get all of the unit tests working in this build, or just a select subset? I'm fine either way, but if the goal is to get everything going, I wonder if there might be some remaining issues. I pulled and then tried building everything, but I'm still seeing a bunch of build errors. I swear that this isn't sarcastic, but: Am I doing something wrong?

I'm ...

1. Running CMake like I always do
2. Opening the GTSAM.sln that pops out
3. Clicking on the little triangle next to Unit tests in Solution Explorer
4. Right-click on check.geometry
5. Selecting Build
6. Seeing a bunch of errors in the console

That process does work for e.g. check.base

I thought maybe the GUI was hiding what I wanted, so I ...

1. Opened the x64 Native Tools Command Prompt for VS 2019
2. cded to my GTSAM build directory
3. Used a spell similar to this workflow file: cmake --build . -j 4 --config Debug --target check.geometry
4. Saw a bunch of errors in the console

Like the GUI-based approach, this does work for check.base.

I saw in one of the commits above that @dellaert 's commit message was fix check.sam, but that's one of the ones that I can't build. This is at 55ad184, btw -- I'm at the current HEAD of develop.

[dellaert]

Sorry, I've been tied up all day.

I get that :-)

In plain English, what's the goal here?

Trying to resolve #91 and #1087, ultimately getting MATLAB and python wrappers working on windows. Borrowing from your little bird analogy, I propose using make check as the canary in the coal mine: this would suss out all compile and linking errors in gtsam proper, make our windows CI better, and move on to any issues in building the MATLAB wrapper. Makes sense?

No, I don't think you're doing anything wrong if it works with base. What are the errors with check.geometry? Maybe paste them in #1087 , as we now closed this PR. I'll repeat this comment there.

[varunagrawal]

My $0.02: I made some great progress on fixing most of the dllexport issues, and only have geometry, nonlinear, sam and slam not passing.

Except for geometry, every other failure is test failure which should be easier to debug than the dllexport issues. I'll create a PR so y'all can take a look.