Considering the rather strange and messy data you seem to be dealing with, I can't image what sort of line breaking would be "appropriate". Tk::Text lets you control how to handle lines that are too wide for the display window: either no wrapping at all (user must scroll the text display left and right to scan all the text), or wrapping at word boundaries (leaving a jagged edge on the side that isn't being "justified"), or wrapping by character (which might be best for you, considering that you have very long stretches with no apparent "word boundary" (i.e. no spaces or punctuation). In any case, "the global record separator" (whatever you are referring to by that) has nothing to do with line breaking in the Text widget.

BTW, PLEASE PUT <code> and </code> AROUND YOUR DATA SAMPLES AND CODE FROM NOW ON. Using any other sort of "tt" or "pre" style markup in your post will tend to really screw up the node display for all of us trying to read your posts.