Santiago Vila <[email protected]> writes:

> I received this from the Debian bug system:

Thanks for forwarding.

> msgunfmt aborts when trying to print some strings containing incomplete
> UTF-8 characters:
>
> $ msgfmt invalid-utf8.po -o invalid-utf8.mo
> $ msgunfmt invalid-utf8.mo
> msgid ""
> msgstr "Content-Type: text/plain; charset=UTF-8\n"
>
> msgid "x"
> Aborted
>
> I would expect that the incomplete character is ignored (with the usual
> "invalid multibyte sequence" warning); or at least that a reason for
> aborting is given.

I'm leaning to displaying some message before aborting, following the
robustness principle[1], where msgfmt should accept incomplete multibyte
sequence, while msgunfmt shouldn't output it in a UTF-8 PO file.  (Note
that MO file is a binary format and does not carry encoding information.)

Footnotes: 
[1]  https://en.wikipedia.org/wiki/Robustness_principle


Reply via email to