#62 Reliable (or loose?) binary file detection

Kimmo Varis

Currently (as of 2.8.0/2.9.0) we have quite a mixed binary file detection algorithms. Quick compare does it differently from full compare. Resulting different results. Time/size methods don't care about binary status.

Earlier binary file detection was essential as we could not show binary files in the file compare. As we now can show all files, there is no such big importance. It is only informal value, not something that modifies WinMerge results and behavior a lot.

Quite a common binary file criteria is that file contains one or more zero bytes. I think that is fine for WinMerge also.

Checking whole file for zero bytes may be slow. But reading few kilobytes only is not reliable always (PDF files are one known problematic case).

We need to determine if we could use the same algorithm for both (full/quick) compare methods. And what kind of algorithm. Reading and checking first / last 10 kilobytes?


Log in to post a comment.