user interface : unexpected order of processing in batch with splitter

[pierrelabreche]

See attached screenshot and workspace script

Observation
Branches can be executed right-to-left, as evidenced by the occurence of breakpoints. The first observed breakpoint occurs in the right branch.

Expectation
The breakpoint in the left branch is expected to occur before the breakpoint in the right branch ( per the usual left-to-right reading convention )

[jsachs]

The order of execution is based solely on the rule that all transformations a transformation depends on must first be recalculated -- otherwise the order (left/right/above/below) does not matter.

[pierrelabreche]

Would it be possible to sort the transformations, such that when a choice of execution order exists, leftmost transformations would be processed first ?

[jsachs]

Of course, but it would be a total rewrite of some complicated code. I don't really see the point.

[pierrelabreche]

The point is not that a big deal now.

The two breakpoints are on a warp transformation.

I was confused by the first breakpoint occuring on the right branch, which I had not noticed since my attention was on the transformation dialog box. I thought there was some sort of corruption in the software as I had not realized that the right branch was where the first breakpoint occured.

It took me a few retrials to realise what was happening.