But as I said, the message shown to the user is wrong. There is no "invalid byte sequence" in the original text. It should say something like "the original text contains symbols that can't be translated into the requested encoding". Giving misleading information during an error is a bug in itself.
The error message is from Glib as your OP said, not Geany, so you need to raise that issue there, but I suspect its probably actually from the underlying iconv or ICU library, don't know what Glib uses on windows. Maybe the error message makes sense at that level where it is converting byte sequences and the byte sequence in the input is invalid in the output?
Geany spends all its time trying to preserve the users data, so its not going to let you save in a way that does not preserve content. Save in an encoding that supports the relevant characters then convert with specialised tools if you want to.
the only other option is to remove those symbols by hand.
Correct, Geany is not a transliteration program, changes can only be made by users.
For what you are trying to do you really need a transliteration program like [uconv](https://linux.die.net/man/1/uconv) (part of [ICU](https://en.wikipedia.org/wiki/International_Components_for_Unicode) IIUC) that can be told what to do with untranslatable data. Adding actions that are rarely used to Geany just complicates it more, and makes more work for the volunteers that support it, better to use the right tool for rare operations. Beat it with a :hammer: not a :screwdriver: :grinning: