Re: [K3d-development] k3dscript engine changes

[Timothy M. S. <ts...@k-...>]

Romain Behar wrote:

 > Posix magic number is a standard
 > for any shell to launch any script, K-3D scripts are
 > standard only in K-3D and #k3dscript magic will never
 > launch K-3D when executed from a shell!

Yes, I understand this ... that doesn't negate the value of having a 
well-defined rule ("the magic token must be at position zero"), versus 
something fuzzier that complicates parsing and may be prone to abuse 
("the magic token ought to be somewhere near the top").

> Both issues came when doing lots of CutN'Pastes from
> an email to a file for replay. I don't see anything
> wrong with allowing such tricks as long as they don't
> bring any drawback.

Problem is, this really complicates parsing, and doesn't catch grossly 
malformed scripts, such as scriptharness.testfailure.k3dscript.001, 
which tries to execute the following:

#k3dscript
Non-existent command

... currently, it succeeds!  While this is a bug that can be fixed, in 
the larger picture it doesn't really solve the problem of cut-n-paste - 
consider a variation on your example:

<command node="/application/document/window/panel1/create_toolbar
/create_PolyCube" command="activate" arguments=""/>

While this will be successfully parsed, it fails because of the newline 
in the "node" attribute.  I would much rather have a straightforward 
syntax that removes ambiguity - "each statement must be on one line".

I also point-out that with the new bug-buddy, we shouldn't have to 
cut-n-paste blackbox logs from e-mail.

Cheers,
Tim