[R-SIG-Mac] quartz hanging

[biii+rsigm@c m@iii@g oii 8p@wexpress@com]

I raised https://github.com/r-lib/xml2/issues/458 and Jeroen was able to 
determine a resolution within the |xml2| package having to do with 
global error handlers and how libxml2 handles errors.

He made a commit 
https://github.com/r-lib/xml2/commit/b05d22a6e72bf94ce58a1483344db477969a9a98. 
After installing that reference, I can no longer trigger the error. I 
don’t know the intended CRAN timeline, but I would expect any CRAN 
version newer than 1.3.8 should have the fix.

Thanks to Stephanie Evert for finding the real culprit! And thanks to 
all others who contributed to the discussion.

Bill

On 7/6/25 13:04, Bill Evans wrote:

> On 7/6/25 11:21, Stephanie Evert wrote:
>
>> I hope this is more readable as a regexp (for me your |-delimited 
>> blocks show up as a single line of text so it's very difficult to 
>> figure out which commands I'm supposed to execute).
>>
>> plot(1:2)
>> # press Cmd-P in quartz window → works
>> library(kableExtra)
>> # press Cmd-P in quartz window → error, hangs, options(error=browser) 
>> doesn't help
>>
>> The first lines were just intended to show that kableExtra is 
>> actually the culprit – which is not quite true. The real cause of the 
>> error seems to be the xml2 package imported by kableExtra.  I get the 
>> same problem with this sequence
>>
>> library(xml2)
>> plot(1:2)
>> # press Cmd-P in quartz window → error, hangs, options(error=browser) 
>> doesn't help
>
> Thank you for verifying and digging deeper! I was making no assumption 
> (yet) that kableExtra was the clear problem; while it does import 
> grDevices and graphics, I wouldn’t expect a simple import to do it. 
> Having said that, xml2 imports neither package directly. In fact, 
> loading each of xml2’s imports does not trigger the error, so that 
> suggests it’s either in xml2’s “./src/*“ files or linking against 
> libxml2.
>
> I apologize for the formatting issue. I use Markdown Here Revival for 
> Thunderbird/Betterbird so that I can type in markdown and have it 
> rendered into HTML. It surprises me only a little that the 
> stripped-out html leaves a bit to be desired.
>
> On 7/6/25 11:44, Jeff Newmiller via R-SIG-Mac wrote:
>
>> The mailing lists strip html formatting... what you see is not what we see unless you set your email client program to compose and send plain text.
>
> That makes sense, I’d forgotten about that.
>
> Bill
>
>> I have reproduced this on MacBook M1Max, macOS Sonoma 14.7.6, R 
>> 4.5.1 RC (2025-06-05 r88292) in a command-line session, so no 
>> Emacs/ESS involved.
>>
>> Best,
>> Stephanie
>>
>>
>>> On 6 Jul 2025, at 01:54, bill+rsigmac using 8pawexpress.com wrote:
>>>
>>> Thank you for trying it, but you did not execute the reprex completely:
>>> you never loaded |kableExtra|. The reprex in-short (I should have
>>> included it this way the first time):
>>>
>>> |plot(1:2) # or ggplot(mtcars, aes(mpg, disp)) + geom_point() # with the
>>> quartz window active, press Cmd-P, no fail library(kableExtra) # with
>>> the quartz window active, press Cmd-P, fail |
>>>
>>
> ​
​
	[[alternative HTML version deleted]]