Re: lynx-dev sidenote on early removal of files passed to viewers

[Henry Nelson]

> But I'd like to know whether there *is* any additional danger involved
> in using a shell script wrapper.  I don't think so, but maybe someone
> who knows better can tell us more.

Thanks very much for the discussion.  It is important to me.  I'm all
ears from comments by others, or second thoughts from you.

>      VIEWER:application/x-some-type:/usr/local/bin/most-and-rm.sh %s
> 
>      with a /usr/local/bin/most-and-rm.sh containing:
> 
>      #! /bin/sh
>      /usr/local/bin/most +s "$1"
>      /bin/rm "$1"

*If* you can somehow find a way to suspend or stop or crash most,
you might be able to grap a shell before the rm is executed???  I'm
no expert, so I tend to be paranoid.  Would a person want that first
line to be "#! /bin/sh -f -e"?

> supposedly and according to lynx.cfg, "most +s" is safe, whatever
> that means exactly.

AFAIK, it means that you cannot call up the default editor or go to
the default shell.

> - Lynx calls VIEWER commands using system(), and system already uses
>   /bin/sh.  On anything resembling Unix, AFAIK.  So whatever problems
>   there are with executing a shell process for interpreting
>   /usr/local/bin/most-and-rm.sh already would exist in case (a).

Except doubled???  System calls /bin/sh, which opens another shell???
Those ??? mean I am totally in the dark, but am suspicious.

> - Assume it doesn't matter to Lynx whether the file exists after the
>   VIEWER command returns.

Doesn't matter to Lynx, but it matters when the maintainer is running
on very tight disk space.

__Henry