rgl causing R and R Studio to abort

[olibee1315]

I am running macOS Big Sur 11.2.2 on a MBP 16-inch, 2019. I recently updated to R 4.0.4 and R Studio 1.4.1103. I have XQuartz 2.8.0_rc2 installed. I have rgl version 0.105.13 installed. Every time I try to load rgl in either R Studio or Terminal the session aborts. Curiously, my desktop, which is running macOS Catalina 10.15.7 21.5-inch, late 2013, R 4.0.3, R Studio 1.4.1103, and XQuartz 2.8.0_rc2 loads and runs rgl as expected.

[dmurdoch]

There are a couple of issues with XQuartz recently, related to its updates. See issue #27. I believe the current rgl devel version works fine: install it using

remotes::install_github("dmurdoch/rgl")

Note that rgl installs tend to be a little tricky, so this might not work for you. In that case, you'll have to wait for the next CRAN release, which I hope will happen within a few days. Unfortunately, there have also been other problems with the XQuartz updates not including a compatible version of libx11 as XQuartz 2.7.11 (see issue #11), so you may need to rebuild rgl to get it to run on your system, or downgrade XQuartz to 2.7.11, which is also kind of tricky.

[olibee1315]

Thank you. Unfortunately this did not solve the problem. I installed the devel version of rgl as suggested and R still aborts. I have already tried downgrading XQuartz; R doesn't abort but the output doesn't show up either. I can still run the process on my iMac just fine, it's just a little slower. Hopefully CRAN will have a fix soon. Thanks again for your help and efforts!

[dmurdoch]

The devel version has just been submitted to CRAN, so unless something new shows up in their tests, waiting for CRAN isn't going to help.

Some tests that might help:

Does x11() work in the places where rgl causes an abort? It should open a graphics window. If it doesn't work, the problem is with XQuartz.

Does sessionInfo() show the rgl version is 0.105.22? Maybe the devel install failed.

If you run quartz() before library(rgl), do you still get an abort? One of the problems that was supposed to be fixed was caused by running quartz() after rgl, so it now tries to run it during the loading process. That might not be happening if it doesn't detect a need for it, but you can force the workaround by setting options(rgl.startQuartz = TRUE) before loading rgl.

[olibee1315]

I have rgl version 1.105.22 installed. x11() does open a graphics window. running quartz() before library(rgl) still causes R to abort. To be clear, the issues I am having that cause R to abort when calling rgl is on a system running macOS Big Sur 11.2.3. All other things the same (i.e., R, R Studio, XQuartz, and rgl) do not cause R to abort on a system running macOS Catalina 10.15.7. Could the issue be with the OS? Thanks again for your efforts!

On Mar 4, 2021, at 11:18 AM, dmurdoch ***@***.***> wrote: The devel version has just been submitted to CRAN, so unless something new shows up in their tests, waiting for CRAN isn't going to help. Some tests that might help: Does x11() work in the places where rgl causes an abort? It should open a graphics window. If it doesn't work, the problem is with XQuartz. Does sessionInfo() show the rgl version is 0.105.22? Maybe the devel install failed. If you run quartz() before library(rgl), do you still get an abort? One of the problems that was supposed to be fixed was caused by running quartz() after rgl, so it now tries to run it during the loading process. That might not be happening if it doesn't detect a need for it, but you can force the workaround by setting options(rgl.startQuartz = TRUE) before loading rgl. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <https://urldefense.com/v3/__https://github.com/dmurdoch/rgl/issues/43*issuecomment-790783906__;Iw!!KwNVnqRv!T24cKo2IVZspVjOqrqKpkMLBbQ-Tde-UE0q6eXpQHhADXTaPl8M2eODyWP_Tl_4aEH_Y$>, or unsubscribe <https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AFJR6LI7R3DXH32NY2TH4OLTB66HFANCNFSM4YTMS5TA__;!!KwNVnqRv!T24cKo2IVZspVjOqrqKpkMLBbQ-Tde-UE0q6eXpQHhADXTaPl8M2eODyWP_Tl9wdYg8G$>.

Brian White, M.S. Aramayo Lab Dept. of Biology Texas A&M University

[dmurdoch]

I don't really know what to suggest, other than the fairly unsatisfactory

remotes::install_github("dmurdoch/rgl", install.opts = "--disable-opengl")

That will build rgl to skip X11 completely. You will be able to see output in a browser using rglwidget() calls, or options(rgl.printRglwidget = TRUE). This is not too bad in RStudio, but very irritating in other front ends.

[olibee1315]

Okay. Like I said, rgl is working as expected on my older system. For what it’s worth, when I issue x11() in R Studio on the machine running macOS Big Sur 11.2.3, I get the following warning: Warning messages: 1: In x11() : X11 protocol error: BadName (named color or font does not exist) 2: In x11() : X11 protocol error: BadFont (invalid Font parameter) 3: In x11() : X11 protocol error: BadFont (invalid Font parameter) 4: In x11() : X11 protocol error: BadFont (invalid Font parameter) 5: In x11() : X11 protocol error: BadCursor (invalid Cursor parameter) I do not get this warning when I issue x11() on the machine running macOS Catalina 10.15.7. Maybe this helps? Best, Brian

On Mar 8, 2021, at 3:50 PM, dmurdoch ***@***.***> wrote: I don't really know what to suggest, other than the fairly unsatisfactory remotes::install_github("dmurdoch/rgl", install.opts = "--disable-opengl") That will build rgl to skip X11 completely. You will be able to see output in a browser using rglwidget() calls, or options(rgl.printRglwidget = TRUE). This is not too bad in RStudio, but very irritating in other front ends. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <https://urldefense.com/v3/__https://github.com/dmurdoch/rgl/issues/43*issuecomment-793105382__;Iw!!KwNVnqRv!SIArGv8AMBGVqx81IzQ3PLZKDJMsttI5vEo70zlAwvFuyMci-hZggKRqYmgfasnSHBJS$>, or unsubscribe <https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AFJR6LI4KA3BHVWIKWS2VLTTCVBBHANCNFSM4YTMS5TA__;!!KwNVnqRv!SIArGv8AMBGVqx81IzQ3PLZKDJMsttI5vEo70zlAwvFuyMci-hZggKRqYmgfaqLoxKSx$>.

Brian White, M.S. Aramayo Lab Dept. of Biology Texas A&M University

[dmurdoch]

That does make it sound like an XQuartz issue, but from the look of the messages, maybe one that can be fixed. It looks something like this one: XQuartz/XQuartz#97

[olibee1315]

That did the trick! It is now working like a charm! Thank you very much for your efforts! Best, Brian

…

Sent from my iPhone

On Mar 9, 2021, at 5:06 AM, dmurdoch ***@***.***> wrote:  That does make it sound like an XQuartz issue, but from the look of the messages, maybe one that can be fixed. It looks something like this one: XQuartz/XQuartz#97 — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[dmurdoch]

Just for the record, could you describe what you did that worked?

[olibee1315]

I uninstalled XQuartz and then removed all existing sub-directories in opt/X11. I then reinstalled XQuartz.

…

Sent from my iPhone

On Mar 9, 2021, at 2:15 PM, dmurdoch ***@***.***> wrote:  Just for the record, could you describe what you did that worked? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.