Re: Hurd translators on FUSE

[olafBuddenhagen]

Hi,

On Fri, Sep 25, 2009 at 08:42:33AM +0200, Arne Babenhauserheide wrote:

> From the Readme: 
> 
> * How To Use
>   Translators linked against libfuse/Hurd don't work exactly like on
>   the Linux operating system. To be more precise: you need to install
>   them using settrans.
> 
>   Some filesystems check for the availibility of a command line
>   argument, which they want to pass to libfuse, and simply fail, if
>   you don't supply it. This is because on Linux you call the
>   translator as a program as such and specify the mount-point as the
>   first argument. 
> 
>   If the fuse-based translator insists on that argument, just provide
>   any useless information (maybe the correct path). This one will not
>   be treated (at least by libfuse). 
> 
>   The filesystem will be put on the node you provide to the settrans
>   command. Anything else is not of any interest.
> 
>   Filesystems that try to export more than one filesystem (I don't
>   know of any example, sorry, but these are generally supported by
>   Linux's FUSE) will fail on the Hurd. This is because of the settrans
>   thing. 

Translators can actually attach themselfs to any node if they wish; so
it would be perfectly possible to make libfuse work exactly like on
Linux I believe...

I guess this would not be the preferred mode of operation for most
modules; but it shouldn't be hard to make them work both as "normal"
translators invoked with settrans, and as traditional programs launched
from the shell like FUSE modules on Linux, thus leaving the choice to
the user.

-antrik-