I'm in the process of releasing version 5.1.3, that will include significant changes to the way character encodings are autodetected. I would like to know if there are any regressions for locales and encodings I could overlooked.
The downloadable tarball is the following:
I plan to release a Windows binary version really soon, probably this evening. This is specially important as Windows has its own set of potential problems regarding character encodings.
I've also written a document describing the new behaviour regarding encodings (the document is also included in the tarball):
If you find any issue, please mail me to email@example.com or subscribe to the Minimum Profit mailing list by writing to firstname.lastname@example.org and say it there.
Update: Here it's the MS Windows binary: