Installation on Ubuntu 24.04 Failure with Mini Racer

[jessequinn]

URL

No response

How to reproduce the issue?

bundle install fails on mini racer

Screenshot(s) or anything else?

No response

[tomhughes]

There was discussion of a similar issue in #5010 that might help, but in general we can't help with problems with third party components - it's something you need to take up with mini_racer.

[gravitystorm]

Re-opening, since we try to support developers with fresh installs, up to the point that they can get the tests running.

@jessequinn I'm sorry to see that it's not working for you. Can you confirm for me (and everyone else who might be able to help better and faster than I can!):

• what architecture you are installing on (e.g. ARM)
• are you using docker or is this running natively on the host
• anything else about your setup that might be different or unusual (e.g. windows-for-linux, anything like that).

[jessequinn]

amd64 image on gcp, natively on 24.04 ubuntu, as per your own install.md, nothing stands out here, typical linux compute engine we have on gcp.

[tomhughes]

I guess the question is does gem install mini_racer work on it's own? If it does then this must be something to do with the combination of modules we install or something and if it doesn't then it seems it's an issue specific to install mini_racer on this platform.

[jessequinn]

[tomhughes]

This is rubyjs/mini_racer#305 / rubyjs/libv8-node#36 which has already been mentioned in #5070 here.

[mmd-osm]

@jessequinn : it would be good if you could post error messages as text rather than screenshot. Search here on Github doesn't work that well with images.

[Hatsune39]

I had the same issue before, check this path:

/var/lib/gems/3.2.0/gems/libv8-node-18.19.0.0-x86_64-linux/vendor/v8/x86_64-linux-gnu/libv8/obj/libv8_monolith.a

If the v8 directory contains x86_64-linux instead of x86_64-linux-gnu, rename the x86_64-linux folder to x86_64-linux-gnu, and the issue should be resolved.

[matkoniecz]

if installing mini racer fails due to bug that they fixed ( rubyjs/mini_racer#305 ) then it can be closed, right? Especially given that mini racer cannot be installed at all, without osm.org code coming in play at all

OSM website is not supposed to patch bugs in mini racer, probably

[tomhughes]

It's not really fixed - they closed the mini_racer bug because they determined the real problem was in libv8 and the PR to fix that is still open (rubyjs/libv8-node#36).

[matkoniecz]

but still, it is not something that would be fixed or fixable by osm.org development?

and in this case

Re-opening, since we try to support developers with fresh installs, up to the point that they can get the tests running.

would be "use different tool or wait for tool to be unbroken", right?