Created attachment 179604 [details] Screenshot showing file differences in the tree, that are not actually different. SUMMARY After a comparison, the tree shows many differences, yet when double clicking on a file, it says it's equal, just not binary equal. I understand the difference, and recognize the Folder > File Comparison Mode > Full analysis option will provide correct diff assessments. But when you have folders you need to compare that are several gigabytes in size and have tens of thousands of files in them (such as a node application with many npm dependencies), waiting an hour or two for a comparison is out of the question. There should be an option somewhere between Binary comparison and Full analysis ... perhaps that does a binary comparison, but then only does a Full analysis on the file differences from the Binary comparison. Or at the very least, in the tree, some kind of visual indicator that a file is not truly different ... only different in binary ... or a filter (similar to Show Identical Files) that can be turned off to hide file differences with equal text. I've been using KDiff3 for over a decade, and I really like it, but In some of my current projects where I need to see the real differences, having a list of binary but not text differences makes the diff almost unusable. STEPS TO REPRODUCE 1. Have two folders where one set has files that have changed in some way, but still have the same text content 2. Run the diff on the folders 3. Double click on a file that claims to be different, but when double clicked, says it's actually the same OBSERVED RESULT The diff tree shows changed files when they are not changed EXPECTED RESULT The diff tree either only showing the truly changed files, or at least showing something that indicates a change is only binary, but the content is not actually different. SOFTWARE/OS VERSIONS Windows: Windows 11 Enterprise
KDiff3 seems to think there is some non-visible difference in the file data. UFT-16 in particular has some situation where the same character can be represented multiple ways. I'll have look at where this message is generated in the code.
Read this again having a third option does seem like a good idea for file comparisons. Not sure what would involved but in should be doable.
Looking the code I don't see the "Full analysis" preference being used in the diff code itself but I do see it affecting the display color. Very confused here.