The following transcript demonstrates the issue.

    $ darcs --version
    1.0.9rc1 (release candidate 1)

    $ darcs --exact-version
    darcs compiled on Nov 12 2006, at 05:43:35
    # configured Mon Oct  9 15:12:44 EDT 2006
    ./configure --no-create --no-recursion

    Context:

    [TAG 1.0.9rc1
    Tommy Pettersson <ptp@lysator.liu.se>**20061008175207]

    $ mkdir x

    $ cd x

    $ darcs init

    $ touch x

    $ darcs rec -lam x x
    Recording changes in "x":

    Finished recording patch 'x'

    $ darcs replace --token-chars '[^ \n\t]' x x x

    darcs failed:  Space not allowed in tokens:  [^ \n\t]

    $ echo $?
    2

On Wed, Dec 20, 2006 at 08:54:58AM +0000, Trent Buck wrote:
>     $ darcs replace --token-chars '[^ \n\t]' x x x
> 
>     darcs failed:  Space not allowed in tokens:  [^ \n\t]

This is wonderful. :-)

I'm sorry, but this in not a bug, although I won't blame anyone
for thinking so. Or it _is_ rather an error message bug. Well,
at the bottom of everything there's a design error, that I've
promised to fix (any year now...)

The word token in the error message should be token specifier,
or something. The whole error message should probably be more
verbose and explain that the reason spaces are not allowed in
token( specifier)s is darcs can't store them with the current
replace patch format.

Hmm, If I'm not misstating the token specifier [^x] would
perfectly well match tokens with spaces in them.

A patch with a well thought out error message would be most
welcome.

See also related issue245

There is a documentation bug hidden in this, in that we actually tell the user: 

You may prefer to define tokens in terms
of delimiting characters instead of allowed characters using a flag such as
\verb!--token-chars '[^ \n\t]'!, which would define a token as being
white-space delimited.

I'll just add a "me too" to this. No other find and replace utility I've ever
used  has required me to also a define a "token spec" in addition to what I find
and replace. When I have tried to do so with darcs I have failed, and thus not
used "darcs replace" when I might have otherwise.  

Could it just be be optional?

On Fri, Jan 18, 2008 at 03:48:20AM -0000, Mark Stosberg wrote:
> I'll just add a "me too" to this. No other find and replace utility I've ever
> used  has required me to also a define a "token spec" in addition to what I find
> and replace. When I have tried to do so with darcs I have failed, and thus not
> used "darcs replace" when I might have otherwise.  
> 
> Could it just be be optional?

The token spec *is* optional.  Perhaps we should document this better?
-- 
David Roundy
Department of Physics
Oregon State University

David Roundy wrote:
> 
> On Fri, Jan 18, 2008 at 03:48:20AM -0000, Mark Stosberg wrote:
>> I'll just add a "me too" to this. No other find and replace utility I've ever
>> used  has required me to also a define a "token spec" in addition to what I find
>> and replace. When I have tried to do so with darcs I have failed, and thus not
>> used "darcs replace" when I might have otherwise.  
>>
>> Could it just be be optional?
> 
> The token spec *is* optional.  Perhaps we should document this better?

Thanks for the follow-up, David.

Is it really optional, if it has a default, and the default isn't what I 
want?

My optional, I meant "the notion is not considered" rather than "a 
default token spec is provided", which I what I believe the correct 
behavior is.

On Fri, Jan 25, 2008 at 08:40:35PM -0000, Mark Stosberg wrote:
> David Roundy wrote:
> > On Fri, Jan 18, 2008 at 03:48:20AM -0000, Mark Stosberg wrote:
> >> I'll just add a "me too" to this. No other find and replace utility I've ever
> >> used  has required me to also a define a "token spec" in addition to what I find
> >> and replace. When I have tried to do so with darcs I have failed, and thus not
> >> used "darcs replace" when I might have otherwise.  
> >>
> >> Could it just be be optional?
> > 
> > The token spec *is* optional.  Perhaps we should document this better?
> 
> Thanks for the follow-up, David.
> 
> Is it really optional, if it has a default, and the default isn't what I 
> want?
> 
> My optional, I meant "the notion is not considered" rather than "a 
> default token spec is provided", which I what I believe the correct 
> behavior is.

I mean that it has a default value.  Actually it has two defaults,
depending on whether or not you have a '.' in either of your tokens.  How
do you imagine it could work without any definition of tokens? The only way
I could imagine doing that would be to construct the set of possible token
characters as a union of the characters in the new and old tokens.  But
that would have the disadvantage that almost any pair of token replace
patches would conflict.
-- 
David Roundy
Department of Physics
Oregon State University

> I mean that it has a default value.  Actually it has two defaults,
> depending on whether or not you have a '.' in either of your tokens.  How
> do you imagine it could work without any definition of tokens? 

When I normally do a find and replace I just tell the system to find "a" 
and replace it with "b". There are no other tokens.

Now, maybe I'm not considering what it means for darcs to implement a 
find-and-replace patch that works reasonably. Maybe token-chars are 
required for the darcs case.

They just aren't for the common text-editor or programming case, which 
is every other case for me.

The good news is that recording a find-and-replace as I'm used to in a 
regular patch has worked perfectly fine for me for as an alternative.

   Mark

On Fri, Jan 25, 2008 at 09:25:53PM -0000, Mark Stosberg wrote:
> When I normally do a find and replace I just tell the system to find "a" 
> and replace it with "b". There are no other tokens.
> 
> Now, maybe I'm not considering what it means for darcs to implement a 
> find-and-replace patch that works reasonably. Maybe token-chars are 
> required for the darcs case.

Yes, darcs only allows invertible changes, which a normal find-and-replace
most certainly is not.
-- 
David Roundy
Department of Physics
Oregon State University

I think this is resolved by my issue245 fix.  Please shout if you disagree.