Large ENVI file IO issues

[DottD]

What is the bug?

I have a ~33GB ENVI file, with dimensions 4,360,048,129 x 2 (hdr attached). gdalinfo gives the following output

Files: large_file
       large_file.hdr
Size is 2, 65080833
Image Structure Metadata:
  INTERLEAVE=PIXEL
Corner Coordinates:
Upper Left  (    0.0,    0.0)
Lower Left  (    0.0,65080833.0)
Upper Right (    2.0,    0.0)
Lower Right (    2.0,65080833.0)
Center      (    1.0,32540416.5)

As you can see, GDAL seems to be able to read only 65,080,833 elements.

Is this a known/intended limitation of the format, or it is a GDAL bug? In the event of an intended limitation, I think that it should be mentioned in the docs.

I have seen that writing such a large file is also a problem with Python, since GDAL expects an int as rows argument (attached a Python script to generate such a large file, which gives the error OverflowError: in method 'Driver_Create', argument 4 of type 'int').

Steps to reproduce the issue

You can use the IDL script attached to generate the sample data.

large_envi_creation.zip

Versions and provenance

OS Version:
Red Hat Enterprise Linux 8.6

Output of gdalinfo --version
GDAL 3.11.3 "Eganville", released 2025/07/12

GDAL installed from conda-forge.

Additional context

No response

[rouault]

Warnings/errors/clamping added in #12783

The GDAL raster data model is limited to signed 32 bits for width, height and number of bands.

Can ENVI open your dataset ? Is that a geospatial dataset ?

[DottD]

Thank you for your prompt support!

ENVI can open the dataset with no issues, probably supports int64 width and height.

My file is, indeed, not a geospatial dataset, it's just a matrix of arcs. There are better data format to store it, but regretfully the software I am using stores it that way.