Re: iconv command problem

[H Xu <xusubsc%gmail.com@localhost>]

On 2011/3/29 2:02, jgw%freeshell.org@localhost wrote:
> H Xu<xusubsc%gmail.com@localhost>  wrote:
>
> > On 2011/3/28 14:19, Marc Balmer wrote:
> > > Am 28.03.11 08:06, schrieb H Xu:
> > > > On both Linux and FreeBSD, the following command could be used to remove
> > > > invalid characters:
> > > >
> > > > $ man cc | iconv -c
> > > >
> > > > However, on NetBSD, it seems that iconv doesn't support this kind of
> > > > usage. Is there any solution?
> > >
> > > Actually it does work like you described.  -c is a valid option to iconv
> > > in NetBSD and it does strip characters that it can not convert.
> >
> > It gives the following on my NetBSD:
> >
> > % man cc | iconv -c
> > Usage:  iconv [-cs] -f<from_code>  -t<to_code>  [file ...]
> >           iconv -f<from_code>  [-cs] [-t<to_code>] [file ...]
> >           iconv -t<to_code>  [-cs] [-f<from_code>] [file ...]
> >           iconv -l
> >
> > Isn't it weird?
>
> Seems you _have_ to specify the -f and -t codes:
>
> ex. ascii to unicode:
>
>         %  man cc>  /tmp/test
>         %  file /tmp/test
>           /tmp/test: ASCII English text, with overstriking
>         %  man cc | iconv -c -f ascii -t unicode>  /tmp/test2
>         %  file /tmp/test2
>           /tmp/test2: Little-endian UTF-16 Unicode English character data, with 
> overstriking
>
> Guess the Linux version makes some assumptions about appropriate defaults?
>
> Cheers,
> Jeff W.

Thanks.

I've tried, the -f and -t values must be specified.

The Linux version uses current locale as default value.

Regards,
Hong Xu
2011/3/29