Latest master build failing

With the release of Pulse 4.0, I’m building the latest on my machine. I’m specifically building commit 87b110bea5425c3c04af4ce66f912dbf5b07ea9b (where the 4.x branch was merged into master), and I get this error:

1>Performing download step (download, verify and extract) for 'Eigen3'
1>-- Downloading...
1>   dst='C:/Users/bob.marinier/git/isaac/pulse-engine/build/External/Eigen3/src/eigen-3.3.9.tar.gz'
1>   timeout='none'
1>-- Using src=''
1>-- [download 100% complete]
1>-- verifying file...
1>       file='C:/Users/bob.marinier/git/isaac/pulse-engine/build/External/Eigen3/src/eigen-3.3.9.tar.gz'
1>-- MD5 hash of
1>    C:/Users/bob.marinier/git/isaac/pulse-engine/build/External/Eigen3/src/eigen-3.3.9.tar.gz
1>  does not match expected value
1>    expected: '609286804b0f79be622ccf7f9ff2b660'
1>      actual: '1452d63b62bcf0798ef8f6ac6cbcf1e8'
1>-- Hash mismatch, removing...

This repeats many times, eventually giving up saying it can’t find the Eigen package.

I built from the 4.x branch successfully just last week. Is there something I should be doing differently?

Pulse pulls and builds Eigen for use by our circuit solver
Somehow, Eigen has lost its repository over the weekend.
They are still working on getting it back up and running
They are providing temporary repositories
So, you can edit this file

Remove the 2 URL lines and replace with these 3 GIT_ lines

add_external_project_ex( Eigen3
  GIT_TAG 3.4.0

Hopefully the usual URL used in our repo will be working again soon…

That worked! Does it matter that pulse was using Eigen version 3.3.9, but with these changes it’s using 3.4.0?

I don’t think it will matter.
I am running our test suite against 3.4 now
Once it all passes (and eigen fixes their repo), I will update our cmake to use 3.4

I can post back here the status later tonight.
But I’m pretty confident that pulse will not be negatively affected by the version update.

You could also pull the 3.3.9 tag if you want

FYI, Eigen fixed their repo
So the build can pull from them again.

I ran our V&V suite against 3.4 and there are a few scenarios that have some mathmatical difference errors, but nothing out of the ordinary. I will need to rebase these scenarios to the new outputs to make the official jump to 3.4, but no need for you to rebuild against 3.3.9 if you are working with the 3.4 build from yesterday

Thanks for the update. I actually intended to build the 4.0 release, so I was on a slightly later commit than I intended. As the official release used 3.3.9, I’ll use that for now.

Privacy Notice