Re: [Chicken-hackers] [patch] utils: qs not escaping pipes

[Peter Bex]

On Mon, Feb 25, 2013 at 02:53:10PM +0100, Michele La Monaca wrote:
> On Mon, Feb 25, 2013 at 9:10 AM, Peter Bex <address@hidden> wrote:
> 
> > Do you have a reference where we can read up on this "caret escaping"?
> 
> I think you already provided one:
> http://technet.microsoft.com/en-us/library/cc723564.aspx

I found its treatment about caret escaping confusing after your
statement that there's a difference between paths and non-paths.
The description from technet seems to indicate it has the same
functionality as \ in UNIX.

> > If it's context-dependent it seems like it would be impossible to decide
> > generally which to use,
> 
> Right. In Windows escape/quoting rules really depends on which command
> is issued. The interpretation of the command-line is mostly up to the
> command itself. And each command has its own interpretation of the
> rules. An example?
> 
> # cd a^ b
> 
> will try to enter the "a b" folder, while
> 
> # md a^ b
>
> will create the folders "a" and "b".

What about these two?

# cd "a b"

and

# md "a b"

> > and we might need an extra option to choose between both quoting forms
> > (abusing the "quoting" option for this is
> > wrong because the difference in Windows is semantical while in Unix it
> > seems to be purely aesthetical).
> 
> Not quite true. For example backslash-escaping has this nice property:
> 
> (qs (string-append s1 s2))  == (string-append (qs s1) (qs s2))

If this is desirable or neccessary, we should just stick with backslash
escaping, if we can make it work properly (that is, without a blacklist).
Having two separate styles of escaping is needlessly complex and
confusing.  Good design involves trade-offs and making decisions, rather
than punting on every decision and forcing it upon the user.  Especially
when it's unclear whether both styles are equally safe!

Cheers,
Peter
-- 
http://www.more-magic.net