Fix build macos

[erikh]

This fixes build issues on my OS X workstation. I had to make several hacks to get this to work, but it's working. :) I do not know if this patch works on linux, as I don't have a good testing environment for it yet.

There is also a small patch here that fixes some errors I saw in test output when failures happened.

[erikh]

Build will likely fail on linux from jenkins reports I've already recieved, but jenkins seems to be down now. I'm going to get a linux environment bootstrapped probably wednesday at which point I'll revisit this patch.

[jainvipin]

@erikh - I can take your diffs and try to make it work on linux environment... and send you back the diffs. Please let me know...

[mapuri]

Not very sure, but I think this might have a side effect in environment where host's go-binary might be in a standard path like /usr/bin.

Since we try to mount the host's environment at this location it might wipe out vagrant-vm's own binary location. So I had conservatively chosen a path unique to the vagrant-vm to avoid landing into such a situation, instead of keeping it same as host's location.

[mapuri]

Looks good.

At a high level I think this change introduces the following:

• In non-Linux environment, it uses vagant-vm's Go environment.
• It allows builds inside the vagrant vm. This gives a way for the user to have a build environment spawned from the workspace. It does needs the workspace to be mounted as read-write inside the vagrant-vm, which concerns me a bit if user were to spawn a multi-node cluster environment and modify the files from the vms. They will need to be slightly careful here, with the possibility of editing files from multiple vms.

Alternatively, I was thinking if user had a linux vm installed on their non-linux host and installed Go there, would they need this still?

[erikh]

That sounds great Vipin; just send a pull request to my fork.

I ordered a linux workstation last night so I can hack on this some more.

-Erik

On Mar 23, 2015, at 11:07 AM, Vipin Jain notifications@github.com wrote:

@erikh https://github.com/erikh - I can take your diffs and try to make it work on linux environment... and send you back the diffs. Please let me know...

—
Reply to this email directly or view it on GitHub #33 (comment).

[erikh]

I tried this. vagrant just halted and caught fire. :)

Your summary is correct, though. The RW mount is not ideal but it is necessary for go install, IIRC. I can try playing with it again in a bit.

-Erik

On Mar 23, 2015, at 11:28 AM, Madhav Puri notifications@github.com wrote:

Looks good.

At a high level I think this change introduces the following:

In non-Linux environment, it uses vagant-vm's Go environment.
It allows builds inside the vagrant vm. This gives a way for the user to have a build environment spawned from the workspace. It does needs the workspace to be mounted as read-write inside the vagrant-vm, which concerns me a bit if user were to spawn a multi-node cluster environment and modify the files from the vms. They will need to be slightly careful here, with the possibility of editing files from multiple vms.
Alternatively, I was thinking if user had a linux vm installed on their non-linux host and installed Go there, would they need this still?

—
Reply to this email directly or view it on GitHub #33 (comment).

[jainvipin]

@erikh - tried out your changes;
Comment1:

• this change is not only not needed, but will not build the netplugin binary appropriately. netplugin binary, which comes from netd.go is in ./ direcotry and not ./netmaster. Unless we move the file locations, let's avoid this change.

[jainvipin]

@erikh - With the three diffs mentioned above, on top of your changes, I was able to pass sanity on a linux system. The diffs were small enough so I didn't send the PR to your fork,

That sounds great Vipin; just send a pull request to my fork. I ordered a linux workstation last night so I can hack on this some more.

[erikh]

NP, I'll fix it up. Not sure how I missed the shell bit.

Thanks Vipin!

-Erik

On Mar 26, 2015, at 1:25 AM, Vipin Jain notifications@github.com wrote:

@erikh https://github.com/erikh - With the three diffs mentioned above, on top of your changes, I was able to pass sanity on a linux system. The diffs were small enough so I didn't send the PR to your fork,

That sounds great Vipin; just send a pull request to my fork. I ordered a linux workstation last night so I can hack on this some more.

—
Reply to this email directly or view it on GitHub #33 (comment).

[erikh]

There seems to be too many problems with this patch, after some thought. I have a linux box now so my investment here is greatly reduced. :)