Re: [Gimp-developer] Cyclic layer-group error when moving linked layers

[Seldom Needy <needthistool gmail com>]

On Fri, Aug 29, 2014 at 1:22 PM, Seldom Needy <needthistool gmail com> wrote:
> I'll need someone else to confirm this in the recent builds to see if it's
> still there, but in GIMP 2.8.8, I've found a small issue (tested on Windows
> XP).
>
> Using the following arrangement of layers and parent-layers (GroupLayers),
> users can make the processor chug, and may experience unexpected behavior.
> Below, the prefix (X) means unlinked, and (L) will mean linked.
>
> __[root image]
> (X)+--------BIG GROUP
> (L)+--------+--------LITTLE GROUP
> (L)+--------+--------+--------LITTLE LAYER ONE
> (L)+--------+--------+--------LITTLE LAYER TWO
> (L)+--------+--------+--------LITTLE LAYER THREE
> (X)+--------+--------BIG BACKGROUND LAYER
> (X)+--------ROOT BACKGROUND LAYER
>
> Upon using the move-tool on any "grabbable" pixel of a linked layer (via
> move's "Pick a Layer or Guide" option), and trying to drag in some
> direction, the linked layers will drift apart or reset to arbitrary
> positions, independent of eachother.
>
> This is contrary to the way that linked layers should behave, and ostensibly
> uses unnecessary processing as well, as program stability and responsiveness
> seems to decrease as this action is performed. This is probably an artifact
> from before GroupLayer was an option, back when all layers in link-mode
> could be handled uniformly.
>
> To resolve this issue, I imagine checking for contained-by and contains
> relationships, and doing more work when the user sets or unsets the "link"
> option for layers would solve the problem. If this operation takes
> noticeable time, it could be done first either when a transformation
> involving the linked layers occurs, or when the dockable Layers-pane looses
> focus following a period of link-toggling by the user there. It might remain
> stored until a delete or link-settings change occurs again.
>
> Regards

For the record, I've added this to the tracker via at
https://bugzilla.gnome.org/show_bug.cgi?id=735906
Report includes a file which should allow the error to be recreated.