Is r-devel-san preferred for current memtest issues on CRAN?

[kylebaron]

Thanks for putting this docker container together. It has been very helpful to me.

Is r-devel-san your preferred setup for reproducing memtest issues on CRAN? It is running for me, but unfortunately I can't reproduce my issue. Not asking for help to troubleshoot that ... just want to make sure I'm running on the best rocker option. I also looked atr-devel-ubsan-clang but can't get that one to run.

Thanks.

[edited]

[eddelbuettel]

It's messy.

As far as I know, the CRAN checks are on some highly customized machines / setups put together by Brian Ripley. On a recent upload of mine, I stated that I was unable to replicate their ASAN / UBSAN issues simply relying on the corresponding R Hub builder -- and Vienna concurred.

Now, @gaborcsardi said he'd have a PR for the Docker container for an update but I have not heard.

Lastly, one other concern is that the regularly scheduled builds on the ASAN/UBSAN containers ... fails each time on a spurious timeout. @cboettig, time to turn those builds off? Thoughts?

[kylebaron]

@eddelbuettel Thanks. Check comes back clean on both rocker/r-devel-san and rhub::check_with_sanitizers.

And thanks for "spamming" R devel back in 2015 and getting all of this started. I wouldn't stand a chance without these tools and all the blog posts providing help.

[eddelbuettel]

Thanks -- meant to mention the blog posts. This all was put together so that we could indeed more easily duplicate what "they" see. But there always is some drift between the setups so ... shrugs ... we can only try to catch up with PRs as needed.

[cboettig]

@eddelbuettel agreed. I've just turned off the cron job triggers for those san & ubsan containers.