Re: [bug-gettext] Gettext incorreclty detects C format

[Bruno Haible]

Hello,

Urmas wrote:
> The strings which have literal '%' sign in the string (like "% done") are
> flagged as c-format, which they obviously aren't. Please fix detection in
> that particular case, as the probability that such combination is a format
> string, is extremely low, and no counter-intuitive no-c-format flag should
> have to be used in that case.     

You already found out about the special comment /* xgettext:no-c-format */,
as documented in the gettext manual [1].

I agree that "% done" does not "look" like a C format string for a human
reader, but it is syntactically a C format string. If it was really meant
as a C format string, and the xgettext program would not mark it as
"c-format", then translators can produce translations that make the program
crash. This is not acceptable. xgettext must not do this. If someone is
to take the responsibility for possible crashes due to missing 'c-format'
annotations, it must be the program's maintainer.

That's why the situation is as it is. xgettext's behaviour will not be changed.

Bruno

http://www.gnu.org/software/gettext/manual/html_node/c_002dformat-Flag.html
-- 
In memoriam Larisa Yudina <http://en.wikipedia.org/wiki/Larisa_Yudina>