Thank you very very much for that information! This problem is the last piece before I can do my first (alpha) release. With xrandr --fbmm I could reproduce the error on the first try so I can do further tests myself now.
And it really is a font issue. The lines in the display are correct but my fonts are not, not even the menu fonts. I will contact the PyQt or Qt community now to find out how to fix that. One already contacted me and notified me of a commandline parameter "-graphicssystem raster" which I have not tried yet. But the person also told me that this is a only workaround. I am excited to finally find a solution for this longstanding problem. Nils On Thu, 19 Jan 2012 20:30:19 +0100 Thomas Jost <schno...@schnouki.net> wrote: > On Thu, 19 Jan 2012 20:12:28 +0100, Thomas Jost <schno...@schnouki.net> wrote: > > On Thu, 19 Jan 2012 19:35:59 +0100, Nils <l...@nilsgey.de> wrote: > > > Hello list! > > > > > > I hope this is not offtopic and I hope to find help here because > > > Archlinux has python3 as default python. > > > > > > I have a software, a Music Notation Editor, here that can start in a > > > one-liner and I need to find a bug that only occurs on some systems. > > > > > > git clone git://github.com/nilsgey/Laborejo.git && cd Laborejo && > > > ./laborejo-qt.sh > > > > > > This will download and run my software Laborejo as normal user without > > > installing anything. The only modifications to your system are new dir > > > .laborejo in your home directory and the downloaded files via git. The > > > only dependency is pyqt and git to download it > > > > > > You will see 5 lines and a symbol. The symbol must be perfectly alingned > > > within the five lines (one pixel above can be tolerated). It should look > > > like this: http://www.wargsang.de/pyqt-bug-report.jpg > > > Do you see that symbol shifted up or down or is it correct? > > > > > > Could you please answer me with the following information attached: Your > > > graphic driver (type ("ati, nvidia, intel" etc. and closed or open > > > source?) and desktop enviroment/window manager (Gnome, KDE, xfce, i3 > > > etc.). If you want to add more information like qt version or X-Server it > > > would be nice as well. Everything display related helps: > > > I believe closed nvidia drivers will shift the symbol. I tested it myself > > > on ati and intel graphics, both 32 and 64 bit and it looked good, both on > > > Linux and Windows. Other users with ati and intel GPU's had no problem. > > > But two persons with an nvidia card had the wrong display. > > > > > > It would be very nice to hear from you! > > > > > > Nils > > > http://www.laborejo.org > > > > > > *The only modifications to your system are new dir .laborejo in your home > > > directory and the downloaded files via git. > > > > Hi Nils, > > > > First, your program looks really good :) A few years ago I tried using > > Lilypond but since I knew nothing about TeX/LaTeX/whatever at that time > > I found it very difficult. > > > > So here's a screenshot of how it looks like on my machine: > Non-text part: image/png > > > > Python 3.2.2, Qt 4.8.0, pyqt 4.9. Using the proprietary nvidia driver > > (290.10), no desktop environment (Awesome WM), xorg-server 1.11.3. > > > > If that can help I could try using the "nouveau" driver too. > > > > Maybe I'm completely wrong, but that could be a screen resolution issue. > > > > AFAIK, most Xorg drivers can detect the screen resolution (in dots per > > inch) correctly. However they then "lie" about it and report 96x96 dpi, > > whatever the real value is. (They do this to mimick Windows behaviour, > > which always consider any screen to be 96x96 dpi so that developers who > > don't understand resolution can still get something to work -- more > > details on https://bugs.freedesktop.org/show_bug.cgi?id=23705). > > > > However, the proprietary nvidia driver does *not* do this: it only > > reports the *real* resolution, and not 96x96. For example right now on > > my laptop: > > > > $ xdpyinfo | grep resolution > > resolution: 129x127 dots per inch > > > > (I discovered this when switching to the "nouveau" driver, which also > > lies about the real resolution -- my fonts looked weird until I added a > > call to xrandr in my .xinitrc to "fix" the reported resolution...) > > > > So if you're using pixel-based placement for a part of your UI and > > point-based placement for other parts (hence depending on the > > resolution), you may very well end with errors like the one I just > > experienced. > > > > You may try playing with "xrandr --fbmm" to change the reported > > resolution of your screen if that can help. Unfortunately it doesn't > > work with the nvidia driver, but again, if you want (and if my > > hypothesis sound reasonable to you) I can try with the "nouveau" driver > > and see if it changes anything. > > > > Hope that helps :) > > > > Regards > > > > -- > > Thomas/Schnouki > > Hi again :) > > Did some more testing using the nouveau driver and playing with "xrandr > --fbmm" to change the Xorg resolution. > > 129x129 (autodetected) in laborejo-129x129.png, 213x267 (random values > :)) in laborejo-213x267.png, 96x96 in laborejo-96x96.png. Now I'm pretty > sure that's the issue :) > > Good luck to solve that! > > Best regards, > > -- > Thomas/Schnouki