Kai Grossjohann <[EMAIL PROTECTED]> writes:
> So it would be useful to allow Tramp to interoperate with really dumb > hosts, which have neither Perl nor uuencode, nor mimencode.
Maybe it's useful to use "sed -ne l"? Please try this on a number of files. Include non-ascii characters in them, put $ signs in them, put tabs and \r in them.
I'm not sure if the output format of "sed -ne l" is the same everywhere. I tried FreeBSD (-current) and GNU/Linux (Debian woody), and they seem to be the same.
The IRIX 6.5.5m version does not conform. Two faults that I've noticed are:
No $ at the end of the line Octal escape of ^A is \01 but it should have three digits \001
There are three versions of sed on Solaris 8. Only the one in /usr/xpg4/bin/ conforms. The ones in /usr/bin/ and /usr/ucb/ behave in the same way as Irix.
AIX 4.3 is okay.
Something to watch out for using standard UNIX utilities for encoding is that the text lines input must not contain NUL (\000) and must not exceed {LINE_MAX} bytes in length.
This would allow us to encode files on the remote end. Lisp code can be used to decode them.
For the other direction, it might work to use "echo -e" and octal escapes. Except that I think that "echo -e" is not standardized. Hm. Is there a program which is known to grok octal escapes everywhere?
Here is a vintage piece of script that has served me well.
# Look for a version of echo that handles escapes if [ `echo "a\062z"` = a2z ] then ECHOE="echo" elif [ `echo -e "a\062z" 2>&1` = a2z ] then ECHOE="echo -e" elif [ `print "a\062z" 2>&1` = a2z ] then ECHOE="print" elif [ `/bin/echo -e "a\062z" 2>&1` = a2z ] then ECHOE="/bin/echo -e" else ECHOE="/bin/echo" fi
-- Pete Forman -./\.- Disclaimer: This post is originated WesternGeco -./\.- by myself and does not represent [EMAIL PROTECTED] -./\.- opinion of Schlumberger, Baker http://petef.port5.com -./\.- Hughes or their divisions.
_______________________________________________ Tramp-devel mailing list [EMAIL PROTECTED] http://mail.nongnu.org/mailman/listinfo/tramp-devel
