Scott Henson <shenson(a)redhat.com> wrote:
On Wed, 15 Jun 2011 23:49:32 -0700, petermity
<6t12-n6js(a)dea.spamcon.org> wrote:
> I updated Cobbler from V2.0.10 to V2.0.11.
>
> Previously, an unescaped undefined $variable anywhere in a
> snippet would cause a snippet to not be included/expanded,
> instead it just left the $SNIPPET line in place.
>
> Now, such a snippet is included/expanded, with any unescaped
> undefined variables just left alone (e.g. "echo $myvar" in the
> snippet remains as "echo $myvar" in the getks output).
>
> Is this intentional? It would seem to mean that I no longer need
> to escape variables, for instance bash variables I'm using in
> %pre or %post, and that password hashes no longer need $ escaped
> or even raw/end raw.
Yes, it is intentional. It applies the same pipeline to snippets as we
do to kickstart templates. Better consistency.
Should you continue to escape variables that you don't want cheetah to
interpret? Yes. It is safer that way. But it will not cause your snippet
to not process anymore.
Thanks for the info and confirmation.
I had become accustomed to the previous picky behavior - it acted
as a sort of "lint" that helped me avoid using variables
improperly in mixups between Cheetah use and kickstart/bash use.
The FAQ and KickstartTemplating pages' mentions of escaping should
be updated for this.
Also I think you could close #587.