#96 Comparison is way too slow.

[konijnpluis]

Hi,

I'm comparing 2 BIG folders, containing lots of 4GB .iso files,
When using Winmerge, it takes approx 30 mins to only compare 2 4GB .iso files,
as seen in the progress window.

When I use the explorer add-on called "HashTab" , and get the results for both these same 2 4GB .iso files,
it does the job in only 2 minutes.......

So please, there needs to be some serious enhancements on the WinMerge comparison algorythm.
Taking such a drastic time to only compare 2 4GB files is way too much.

[konijnpluis]

Edit: I forgot to be specific about the comparison method used by HashTab, it's using CRC32/MD5/SHA-1 .... I wonder why the folder comparison method takes such a long time for only comparing 2 items, while HashTab is able to calculate SHA-1 checksums for the same 2 files in just 2 minutes........

Or isn't the "full contents" methods from WinMerge a simple CRC32/MD5/SHA-1 checksum for .ISO files ?????

[Matthias]

in full contents it's comparing byte by byte.

[konijnpluis]

matthias,

I understand that it's byte-by-byte, but why in a folder comparison ???????

It would be logical to be byte-by-byte when you do a file compare, but why for FOLDER COMPARISON ?

When a folder comparison finds 2 non-identical files, you can compare them byte-by-byte in the following file compare, but I don't understand why a folder compare would use that method instead of a checksum comparison ??????

[Kimmo Varis]

Even using checksum is waste of time for some people. Just compare the time and/or size and that's it.

You clearly assume all files are binary files which isn't case for lots of people. And when you have text files then you definitely want to ignore for example EOL differences. Likewise it is good to check if there is BOM bytes or not. Different Unicode encodings can be handled also.

It is true we could offer checksum compare and indeed if somebody sends a patch I'm happy to apply it. But checksum compare is only solution for certain type of compares.

And yes, there could be some logic that we ignore certain checks for large files. But then who decides which file is large enough? There certainly are people having huge text files. And the circle starts from the beginning...

[konijnpluis]

Thanks a lot for considering it kimmov,

It would be a serious step forward in usability for the users, if there was, for example, a much more effective way if choosing your comparison method if we didn't have to go trough the options window but if there was a direct "Comparison Method" item on the top toolbar, If there would be an item like "MD5 checksum" that would greatly increase the usability and speed of WinMerge.