Comment #3 on issue 2159 by [email protected]: Patch: lexer.ll: Warn about
non-UTF-8 characters
http://code.google.com/p/lilypond/issues/detail?id=2159
Ok, I am not satisfied. Encode the following in latin-1 and check the
result after applying the latest patch:
%{ "Dies ist ein schööner und ergötzlicher Test" %}
There is a wagonload of programming errors. Why? Because the error
messages and other stuff want to determine the column numbers for the error
messages, and in the absence of legal utf8, there _are_ no well-defined
column numbers. And the code supposed to calculate them is not robust
enough to do something sensible.
So since Lilypond does not seem robust against sensibly dealing with
non-UTF-8 input elsewhere, I suggest making this a _fatal_ error, and if
the fatal error is thrown strictly _before_ the first bad byte, it still
has a valid column number itself.
A transparent garbage-in-garbage-out can be dealt with warnings, but
garbage-in-failure-out should be an error.
_______________________________________________
bug-lilypond mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/bug-lilypond
