On Wed, 05 Jan 2005 19:04:45 -0500, Ales Hvezda <[EMAIL PROTECTED]> wrote: > [I wrote:] > >One more data point: when I was reading in my old schematics with > >20041228 on OS X, the initial view sometimes showed a portion of a > >component, but I did not see any nets. It was difficult to reproduce. > > > >Did some of the redraw code change? > > > > I haven't seen any of these bugs. I didn't change any redraw > code from 20040111 to 20041228.
I wonder if this has to do with the changes in where the paper size stuff is initialized. (On a side topic, how do you use gdb on noweb source? When I try and run gschem from the src directory, it picks up the proper "../noweb/blah.nw" path, but the line numbers are off, making it next to impossible to use breakpoints or single-stepping.) Another interesting data point: In 20040111, if I place a single component in the upper-right corner of the window (new document), there is an invisible box (in world coordinates) that components seem to be clipped to when the code determines what the page extents are. In this case, the upper right corner coordinates are roughly (121000, 76000) Same thing in 20041228, except that the box only extends to (24000,17600) In both cases, if the component is mostly outside this box, it will not be shown during 'zoom extents'. 20040111 is more lenient about zooming way out (compare the coordinates), which explains the apparent regression when opening old schematics. Are any Linux users seeing this issue in either version? I can reproduce the invisible box problem with 20040111 (same coordinates as above) using Debian packages, but I haven't built 20041228 from source on Linux yet. -- - Charles Lepple
