Re: [Bug-ed] Global substitutions that insert newlines

[Antonio Diaz Diaz]

Rob Arthan wrote:
> This is an annoying side-effect of the POSIX standardisation: some
> implementations of ed allowed the terminating character to be dropped from
> a substitution command (i.e., they allowed "s/<pattern>/<text>" as well as
> "s/<pattern>/<text>/").  This introduced an "ambiguity" (in the POSIX
> description and not, as far as I know, in any implementation) that
> eventually caused a useful feature to be dropped from POSIX. A better
> solution in the POSiX standard would have been either always to require
> the terminating character (as the ed implementations I grew up with did)
> or to legislate in favour of the newline interpretation if the terminator
> was omitted in the ambiguous cases (which is what the GNU implementation
> originally did).

I agree with you, and I think that the line-splitting functionality 
could perhaps be recovered (and POSIX fixed) given that GNU ed already 
requires the last delimiter to be present in this case:

http://www.gnu.org/software/ed/manual/ed_manual.html#Commands

"The last delimiter can't be omitted if the 's' command is part of a 'g' 
or 'v' command-list and is not the last command in the list, because the 
meaning of the following escaped newline becomes ambiguous."

I'll look into this ASAP (but not very soon).


Best regards,
Antonio.
--
If Spain were not an authoritarian State, catalans would have been 
allowed to hold their independence referendum years ago. Quebec and 
Scotland had their independence referendums. Why can't Catalonia have one?