On Wed, Dec 19, 2007 at 06:40:43PM -0500, DJ Delorie wrote: > > Ben, as far as I can tell this is yours. It broke on the 2nd-ish. > The LayerN patch sounds familiar (we've had this bug before).
I'll check into it, but the LayerN->max_layer change was to fix a bug where buffers had LayerN of 0. For the main board, my change just moves from Data->LayerN (where Data is PCB->Data) to max_layer, which is PCB->Data->LayerN, so there should be no change. I've seen problems in the past where I changed layer grouping and pads no longer cleared surface polygons. I never figured out why it happened. > revision 1.122 > date: 2007/12/02 08:44:52; author: bjj; state: Exp; lines: +10 -4 > Partial fix? [ 1775101 ] Undo with the line tool sometimes switches to wrong > node That can't be it... > revision 1.68 > date: 2007/12/02 09:35:40; author: bjj; state: Exp; lines: +3 -3 > Data->LayerN should generally be avoided. It's not initialized for buffers. This is the change in misc.c that I made specifically for the patch. This is the bug that prevented the bounding boxes of buffers from being computed if they didn't contain a via or other "all layer" object. They were "empty" because LayerN was 0 and things like lines could not be found. This is the bug that kept groups from moving up/left at start (and lots of more subtle bugs once you know what to look for). > revision 1.59 > date: 2007/12/02 09:35:40; author: bjj; state: Exp; lines: +9 -9 > Data->LayerN should generally be avoided. It's not initialized for buffers. This change to the polygon code was an attempt to fix potential unknown bugs due to uninitialized LayerN in buffers. This is the only change I'd consider risky. If I can reproduce the bug, the polygon.c change is the one I'd revert first. However, I still think that the ultimate LayerN in the "main board" case is the same as it was before (not just numerically, but literally the same variable, just accessed differently). -- Ben Jackson AD7GD <[EMAIL PROTECTED]> http://www.ben.com/ _______________________________________________ geda-dev mailing list [email protected] http://www.seul.org/cgi-bin/mailman/listinfo/geda-dev
