Vagrant / SITL build totally broken

Has anyone tried to setup SITL through Vagrant on Windows lately ?

The documentation on http://ardupilot.org/dev/docs/setting-up-sitl-using-vagrant.html is certaintly not up to date anymore (sim_vehicle.sh was deleted) and it seems its “replacement” just doesn’t build anything anymore (various commands that are not found).

I did a clean check out of the repository with all git submodules updated, so this should work out of the box, but it simply doesn’t …

It seems an upgrade was made from Ubuntu Trusty to Wiley lately. Anyway, I would like to hear if other people succeed in following the steps from the documentation (from a fresh repo of course !) … or any other steps.

Also, from a developer perspective: wouldn’t it be a better idea to let users check out a certain “release” branch from the repo ? It seems that the “master” is often in a “broken” state, due to the lack of time to test added features or fixes.

It not being maintained. If you do fix it, please feel free to lets us know and we can include the updated instructions

Otherwise install a VM with Ubuntu Linux on it and follow these instructions http://ardupilot.org/dev/docs/setting-up-sitl-on-linux.html#setting-up-sitl-on-linux

Bill: I am kind-of-maintaining it.

My email reply to the original post seems to be taking its time arriving…

My email, since it seems to have been lost:

On Thu, 8 Sep 2016, Benedikt Naessens wrote:

Has anyone tried to setup SITL through Vagrant on Windows lately ?

Sorry, no, only starting Vagrant on an Ubuntu host.

The documentation on
Setting up SITL using Vagrant — Dev documentation is
certaintly not up to date anymore (sim_vehicle.sh was deleted) and it
seems

Ouch. I think you found the one remainging instance of sim_vehiclee.sh!

You can ignore that section; nothing needs to be done using sudo there.

its “replacement” just doesn’t build anything anymore (various commands
that
are not found).

OK, that’s a problem. I’ve just done a “vagrant up” myself, and it all
seems to wortk. Could you peek back through the logs and see if there was a
network error while it was attempting to install packages, please? I’m
wondering if we can improve out error detection somehow.

I did a clean check out of the repository with all git submodules updated,
so this should work out of the box, but it simply doesn’t …

OK, my host-machine setup certainly isn’t “out of the box”. I’ll install
Vagrant on my windows machine here and try an “up” and see what I see.

Also, from a developer perspective: wouldn’t it be a better idea to let
users check out a certain “release” branch from the repo ? It seems that
the
“master” is often in a “broken” state, due to the lack of time to test
added
features or fixes.

We don’t usually recommend users fly master. If you’re developing, you
should :slight_smile:

You can always do a git-checkout based on a release branch (e.g. Randy’s
3.4-rc tags); those are the closest we have to a “release” branch to my
thinking.

That’s great news, go @peterbarker

@benedikt.naessens When you installed git for windows, did you specify that Unix-style line ending should be used all the time?

If not, you will get an error like, " : Command not found" - which isn’t particularly useful.

Also note that if you’ve done a build on the host machine you will need to remove the build directory before building.

Does any of that solve your problem?

If note, please paste in some error text for me to digest :slight_smile:

I now have sim_vehicle.py running on Ubuntu on Windows, however I am now getting “time moving backwards” issues - probably because I need the VirtualBox tools installed on the VM or somesuch.

Peter

Peter,

It was indeed the Unix style line ending that did it for me … my mistake ! Anyway, I will do a documentation fork and try to update the aformentioned documentation page this weekend.

Thanks !

Hello Bendikt/Peterbarker: can you please help as I am working on similar issue and reported with all logs here.

Thanks,

updated: 14-sep-2017: finally fix here.