gscan2pdf / Bugs / #439 Test-suite fails with ImageMagick >=7.1.2-9

Andreas Wiese - 2025-12-19

Addendum: I'm trying to build gscan2pdf-2.13.5 with perl-5.40.0 on NixOS 25.11.

Last edit: Andreas Wiese 2025-12-19

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Jeffrey Ratcliffe - 2025-12-20

Thanks for the report. Debian testing still has 7.1.2-8, so I can't reproduce this. Please try the following with 7.1.2-9 and attach the jpg to the report:

magick -units Undefined -density 300 xc:white test.jpg

What does the following then return?

identify -verbose test.jpg

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

$ magick -version
Version: ImageMagick 7.1.2-9 Q16-HDRI x86_64 5cff2fc4f:20251128 https://imagemagick.org
Copyright: (C) 1999 ImageMagick Studio LLC
License: https://imagemagick.org/script/license.php
Features: Cipher DPC HDRI OpenMP(4.5) 
Delegates (built-in): bzlib cairo djvu fftw fontconfig freetype heic jng jp2 jpeg jxl lcms lqr openexr pangocairo png raqm raw rsvg tiff webp x xml zlib zstd
Compiler: gcc (14.3)
$ magick -units Undefined -density 300 xc:white test.jpg
$ identify -verbose test.jpg
Image:
  Filename: test.jpg
  Permissions: rw-r--r--
  Format: JPEG (Joint Photographic Experts Group JFIF format)
  Mime type: image/jpeg
  Class: PseudoClass
  Geometry: 1x1+0+0
  Units: Undefined
  Colorspace: Gray
  Type: Grayscale
  Base type: Undefined
  Endianness: Undefined
  Depth: 12/16-bit
  Channels: 2.0
  Channel depth:
    Gray: 16-bit
  Channel statistics:
    Pixels: 1
    Gray:
      min: 4095  (1)
      max: 4095 (1)
      mean: 4095 (1)
      median: 4095 (1)
      standard deviation: 0 (0)
      kurtosis: 0
      skewness: 0
      entropy: 0
  Colormap entries: 4096
  Colormap:
  Rendering intent: Undefined
  Gamma: 0.454545
  Matte color: grey74
  Background color: white
  Border color: srgb(223,223,223)
  Transparent color: black
  Interlace: None
  Intensity: Undefined
  Compose: Over
  Page geometry: 1x1+0+0
  Dispose: Undefined
  Iterations: 0
  Compression: JPEG
  Quality: 92
  Orientation: Undefined
  Properties:
    date:create: 2025-12-20T18:23:06+00:00
    date:modify: 2025-12-20T18:23:06+00:00
    date:timestamp: 2025-12-20T18:23:12+00:00
    jpeg:colorspace: 1
    jpeg:sampling-factor: 1x1
    signature: ea0eb021dc2ad082eb7325b55ba9a512899156fc727e9c318c8ee581965fd5b0
  Artifacts:
    verbose: true
  Tainted: False
  Filesize: 160B
  Number pixels: 1
  Pixel cache type: Memory
  Pixels per second: 4404P
  User time: 0.000u
  Elapsed time: 0:01.000
  Version: ImageMagick 7.1.2-9 Q16-HDRI x86_64 5cff2fc4f:20251128 https://imagemagick.org
$

test.jpg

Jeffrey Ratcliffe - 2025-12-21

If you compare the output of identify, whereas in previous versions, the resolution metadata was still present in the image, plus the undefined units, in this new version, the resolution is completely missing. I'm calling this a bug in Imagemagick.

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Andreas Wiese - 2025-12-21

Okay, so I filed a bug against ImageMagick (I hope to have gotten this right). But I'm not quite convinced here. So yes, identify doesn't report a resolution anymore, but isn't this actually correct without a unit? I mean, what does 300x300 tell without a unit?

Edit:
Okay, the bug report leading to the breaking commit explains this for agnostic people like me:

From the JFIF specification:

units: 1 byte

0 = no units, X and Y specify the pixel aspect ratio

1 = X and Y are dots per inch

2 = X and Y are dots per cm

When units=0, the density values are explicitly defined as aspect ratio, not resolution.

Last edit: Andreas Wiese 2025-12-21
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Jeffrey Ratcliffe - 2025-12-22

300x300 without a unit should normally default to dots (pixels) per inch. And living in a land that uses SI units, I would be also happy for it to default to pixels per cm. But the new behaviour ends up being 72 ppi, which makes far less sense than anything else.

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Jeffrey Ratcliffe - 2025-12-31

This bug has now hit Debian sid. I'll have to skip the test in sid to avoid FTBFS. Your bug report is now seeing some attention.

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

Test-suite fails with ImageMagick >=7.1.2-9

Group

Searches

Help

#439 Test-suite fails with ImageMagick >=7.1.2-9

Discussion