[cvsnt] Re: Binary vs text guesser

[Tony Hoyle]

Terris Linenbach wrote:

> DLLs.  This would make script-driven imports actually feasible.  Even
> better if the logic could determine whether a file is UNICODE, either by
> looking for the UNICODE header bytes, or via wildcards.
> 

Unicode is automatic anyway, provided there's a correct header.  The -ku
only specifies the output format (since Unicode is internally translated to
UTF8).

Binarry is set by the cvswrappers file which seems to me the best way of
doing it - the frontends just add a client side safety net.  Scripted
imports work fine if this file is set.

Tony