[Bioc-devel] Need help figuring out GeometryDoesNotContainImage-error on machv2-build for chimeraviz

[Hervé Pagès]

Ok so I'm changing my mind about this. I suspect that the error is 
actually related to the warning. The error comes from the magick package 
(a wrapper around the ImageMagick software) and it indicates a failure 
to crop an empty image. It can easily be reproduced with:

   ## Generate an empty image.
   png("myplot.png", bg="transparent")
   plot.new()
   dev.off()

   ## Try to crop it.
   magick::image_trim(magick::image_read("myplot.png"))
   # Error in magick_image_trim(image, fuzz) :
   #   R: GeometryDoesNotContainImage `/Users/biocbuild/myplot.png' @ # 
warning/attribute.c/GetImageBoundingBox/247

So I suspect that what happens is that the images generated on Mac by 
the code in the vignette are empty (because of the semi-transparency 
problem on Mac) which would explain why later knitr fails to crop them 
(it uses magick::image_trim() for that).

I don't exactly understand why we wouldn't have seen the problem on 
merida1 though (same version of knitr (1.28) and magick (2.3) on both 
machines) but it seems that chimeraviz has changed significantly between 
BioC 3.10 and 3.11. Did you start using semi-transparency recently in 
your plots?

Best,
H.


On 4/23/20 19:42, Hervé Pagès wrote:
> Hi Stian,
> 
> I went on machv2 and gave this a shot. I can reproduce the 
> GeometryDoesNotContainImage error in an interactive session. I don't 
> have an answer yet but I was curious about the "semi-transparency is not 
> supported on this device" warning and was wondering if it could somehow 
> be related with the error.
> 
> It turns out that the warning is actually easy to reproduce on a Mac 
> with something like this:
> 
>    plot.new()
>    lines(c(0.1, 0.22), c(0.5, 0.44), type = "l", lwd = 1, col = 
> "#FF000080")
> 
> I think that the 4th byte (80) in the color specification ("#FF000080") 
> is the level of transparency.
> 
> I can get this warning on machv2 **and** merida1. Some googling 
> indicates that this is a pretty common warning on Mac. Since we don't 
> get the vignette error on merida1 I think it's unlikely that the warning 
> is related to the error.
> 
> I'll keep investigating the GeometryDoesNotContainImage error...
> 
> H.
> 
> 
> On 4/22/20 01:59, Stian Lågstad wrote:
>> I'm still unable to reproduce this error on my end. If anyone with a mac
>> could try building locally I would be very grateful. Thanks.
>>
>> On Sat, Apr 18, 2020 at 4:06 PM Stian Lågstad <stianlagstad using gmail.com>
>> wrote:
>>
>>> Hi,
>>>
>>> I'm haven't been able to figure out this error for the latest machv2 
>>> build
>>> for chimeraviz:
>>>
>>> ```
>>> Warning in doTryCatch(return(expr), name, parentenv, handler) :
>>>    semi-transparency is not supported on this device: reported only once
>>> per page
>>> Quitting from lines 108-126 (chimeraviz-vignette.Rmd)
>>> Error: processing vignette 'chimeraviz-vignette.Rmd' failed with
>>> diagnostics:
>>> R: GeometryDoesNotContainImage
>>> `/private/tmp/RtmpdBrrvk/Rbuild9ed5154894cc/chimeraviz/vignettes/chimeraviz-vignette_files/figure-html/unnamed-chunk-7-1.png' 
>>>
>>> @ warning/attribute.c/GetImageBoundingBox/247
>>> --- failed re-building ‘chimeraviz-vignette.Rmd’
>>> ```
>>>
>>> The build in question:
>>> https://urldefense.proofpoint.com/v2/url?u=http-3A__bioconductor.org_checkResults_3.11_bioc-2DLATEST_chimeraviz_machv2-2Dbuildsrc.html&d=DwIFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=BK7q3XeAvimeWdGbWY_wJYbW0WYiZvSXAJJKaaPhzWA&m=Q1W5ctHOQ2Hjw49pX7VAGBn7-u5l3mAqH1rt9tFNnZM&s=A7jGEFF5ehz2Hsxmlr_vGmPSX3Xy2SwZErgyi1mPIuw&e= 
>>>
>>>
>>> If anyone has seen something like this before then I'd appreciate some
>>> help. Thank you!
>>>
>>> -- 
>>> Stian Lågstad
>>> +47 41 80 80 25
>>>
>>
>>
> 

-- 
Hervé Pagès

Program in Computational Biology
Division of Public Health Sciences
Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N, M1-B514
P.O. Box 19024
Seattle, WA 98109-1024

E-mail: hpages using fredhutch.org
Phone:  (206) 667-5791
Fax:    (206) 667-1319