On 11/6/2014 9:58 AM, Hans Hagen wrote:
> we could have a specific filename cleaner but of course there will be
> users who have paths that have spaces after the \ ... okay, we could
> just not support evil paths like that
>
> Hans

A rigorous filename cleaner would have to treat a huge number of possible
Unicodes. This is overkill. I believe the best would be to have a cleaner
for all ascii "symbols". That would fit most purposes of writing file names
and URIs:

  !#"$%&'*()+,-./:;<=>?@[\]^_`{|}~

Of the above characters the following are disallowed on NTFS:

  "*/:<>?|

Still I think they should be included in the cleaner. If % or # pose
problems then they may be omitted.

With the current "absolute"-cleaner, a space after \ is preserved
correctly, so I don't understand your comment:

\definetype[ctype][compact=absolute]\starttext
\bTABLE\bTR\bTD \ctype{\a\ b} \eTD\eTR\eTABLE\stoptext

I would be glad if you could implement such a "ascii-symbol"-cleaner.

Thanks,

Christoph