Using Zsh on a embedded system to configure GPIOs

Using Zsh on a embedded system to configure GPIOs

[meino.cramer]

Hi,

on the Arietta G25 embedded Linux system GPIOs are configured
via echo-commands to files below /sys/...

To prevent overwriting regular files with redirection commands
I configured zsh to hit on my fingers when trying to do so. To
overwrite files I need to use ">!" instead of ">".
Which per se is a GREAT feature, it becomes annoying when manipulating
the contents of that virtual files below /sys (and /proc and others).

Is there any way to automate this feature NOT to hit on my fingers,
if I want to write to files below /sys, /proc,, /dev...or
(better) to recognize, if such a file is a virtual one?

Thank you very much in advance for any help!

Best regards,
Meino

Re: Using Zsh on a embedded system to configure GPIOs

[Bart Schaefer]

On Nov 29, 10:10am, meino.cramer@gmx.de wrote:
}
} To overwrite files I need to use ">!" instead of ">".
} 
} Is there any way to automate this feature NOT to hit on my fingers,
} if I want to write to files below /sys, /proc,, /dev...or
} (better) to recognize, if such a file is a virtual one?

I confess to being confused by the phrase "to hit on my fingers" ...
but I think I can answer anyway.

The HIST_ALLOW_CLOBBER option replaces ">" with ">|" in the history,
so that if you forget to use the clobbering form all you need to do
is recall the previous command with "!!" or the line editor.  Does
that help?

NO_CLOBBER already behaves differently for non-regular files, so that
you can do things like "print foo > /dev/pts/3" without having to use
">|" (though that won't work for /dev/tty for a different reason).

So if there is some way to identify a "virtual file" e.g. with stat(),
that could potentially also be handled.  However, I think virtual files
masquerade pretty thoroughly as real ones except for the fact that they
appear to have size 0 yet contents can still be read.

I suppose that, since the point of NO_CLOBBER is to prevent destroying
existing file contents, redirection could be allowed to succeed for
existing zero-sized files, which would then work for virtual files.

However, that's not the way it works now, and zsh-workers would need to
have some discussion of whether that change is acceptable.

Re: Using Zsh on a embedded system to configure GPIOs

[Mikael Magnusson]

On Sat, Nov 29, 2014 at 8:35 PM, Bart Schaefer
<schaefer@brasslantern.com> wrote:
> On Nov 29, 10:10am, meino.cramer@gmx.de wrote:
> }
> } To overwrite files I need to use ">!" instead of ">".
> }
> } Is there any way to automate this feature NOT to hit on my fingers,
> } if I want to write to files below /sys, /proc,, /dev...or
> } (better) to recognize, if such a file is a virtual one?
>
> I confess to being confused by the phrase "to hit on my fingers" ...
> but I think I can answer anyway.
>
> The HIST_ALLOW_CLOBBER option replaces ">" with ">|" in the history,
> so that if you forget to use the clobbering form all you need to do
> is recall the previous command with "!!" or the line editor.  Does
> that help?
>
> NO_CLOBBER already behaves differently for non-regular files, so that
> you can do things like "print foo > /dev/pts/3" without having to use
> ">|" (though that won't work for /dev/tty for a different reason).
>
> So if there is some way to identify a "virtual file" e.g. with stat(),
> that could potentially also be handled.  However, I think virtual files
> masquerade pretty thoroughly as real ones except for the fact that they
> appear to have size 0 yet contents can still be read.
>
> I suppose that, since the point of NO_CLOBBER is to prevent destroying
> existing file contents, redirection could be allowed to succeed for
> existing zero-sized files, which would then work for virtual files.
>
> However, that's not the way it works now, and zsh-workers would need to
> have some discussion of whether that change is acceptable.

Maybe we could add a NO_DESTRUCTIVE_CLOBBER which allows >> to
nonexisting files, > to zero sized files and possible etceteras, but
not > to nonzero files? I personally never find `>> nonexist` aborting
to be useful.

-- 
Mikael Magnusson