Hi Phil (et al), sorry for all the site issues! I just installed Exiftool 13.59 and started to see timeouts when using Exiftool to copy exif from one image to another (when resizing). Here is the brief summary from Claude:
ExifTool 13.56 introduced a performance regression in GetValue: on each tag lookup that misses, it now scans the entire tag set case-insensitively. On images with very large XMP (e.g. heavy Lightroom masking/retouch data with thousands of entries), this becomes O(N²), turning a sub-second metadata write into ~18 seconds.
I have been able to manually recreate this with something like:
time exiftool -tagsFromFile @ -Comment -o - image.jpg > /dev/null
I'll send over a sample image directly that is easy to recreate with.
Thanks,
Lee
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Thanks. Yes, I worried about the possible performance hit of this change. I'll have to think about the best way around this. And I'll have to remember exactly why this patch was applied. (There was something not working properly with some Composite tags if I recall correctly.)
I got the test file, thanks.
Phil
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Phil (et al), sorry for all the site issues! I just installed Exiftool 13.59 and started to see timeouts when using Exiftool to copy exif from one image to another (when resizing). Here is the brief summary from Claude:
ExifTool 13.56 introduced a performance regression in GetValue: on each tag lookup that misses, it now scans the entire tag set case-insensitively. On images with very large XMP (e.g. heavy Lightroom masking/retouch data with thousands of entries), this becomes O(N²), turning a sub-second metadata write into ~18 seconds.
I have been able to manually recreate this with something like:
time exiftool -tagsFromFile @ -Comment -o - image.jpg > /dev/null
I'll send over a sample image directly that is easy to recreate with.
Thanks,
Lee
Hi Lee,
Thanks. Yes, I worried about the possible performance hit of this change. I'll have to think about the best way around this. And I'll have to remember exactly why this patch was applied. (There was something not working properly with some Composite tags if I recall correctly.)
I got the test file, thanks.