Skip to content
S.-H. Dan Shim edited this page May 3, 2021 · 34 revisions

Problems and fixes

2021/05/01

  • The instruction on installation was updated.

2020/02/19

  • Segmentation fault 11 issue by resizing the window come from tk. I tested up to version tk 8.6.8 and they are all fine. The problem comes from tk 8.6.9. It seems that tk 8.6.10 may not have this issue anymore

  • qdarkstyle 2.8 has an issue with resizing the shortcut pushbutton icons at the top. For now stay with 2.7. I tested this issue with newer version pyqt 5.12.3. But it did not fix the problem.

2019/08/10

  • Segmentation fault 11 issue by resizing the window come from tk. I tested up to version 8.6.8 and they are all fine. The problem comes from version 8.6.9.

2019/07/23

  • matplotlib 3.1 seems to break down the helvetica font support for pdf output.

    • I found a solution for this. This is because of the problem in ~/.matplotlib/fontlist-310.json. Make sure the folder under ./matplotlib/mpl-data/fonts/ttf is linked to Helvetica font information in the json file.
  • numpy 1.6.x seems to have problem in mac installation. I had to install using pip when it does not work.

2019/07/22

  • Some combination of incompatible variants found for the conda environment for PeakPo

    • found from Huawei Chen's version (very old): dill 0.2.8.2, pyfai 0.14.2, I will call it peakpo2017.
    • found from Shim (env=peakpo7721): dill 0.2.6, pyfai 0.14.2, I will call it peakpo2018.
    • found from Shim (env=peakpo7721-new): dill 0.2.9 or later, pyfai 0.17.0 or later, I will call it peakpo2019.
    • fabio version does not seem to matter.
  • To enable PeakPo dpp files in jupyter, I need to have pyzmq version higher than 17 and tornado higher than 5. This is not compatible with PyQt requirement for PeakPo. So I will create j-peakpo201x environments for those.

  • Seg fault 11 issue in Mac

    • qdarkstyle causes crash if version is higher than: 2.7
    • pyqt should be version 5.6.0, not higher.
    • qt should be version 5.6.2, not higher.

2019/04/12

  • The dpp compatibility issue is from pyfai. pyfai has a change between version 0.15.0 and 0.17.0, which makes the problem. So if you have an old dpp you cannot read, do the following.

    conda create --name <env-name-for-old-pyfai> --clone <your-current-peakpo-environment-name>
    
    conda activate <env-name-for-old-pyfai>
    
    conda install pyfai==0.15
    
  • In conclusion, you just need to have two environments, one with pyfai 0.15.0 and the other with pyfai 0.17.0 or higher.

2019/03/11

  • Versions of dill, pyfai, and fabio are problem for now.

  • If you have an old dpp file, use dill==0.2.6, fabio=0.4.0, and pyfai=0.14.2.

    conda install dill==0.2.6 # This forces installation of particular version in your conda environment.
    

2019/02/15

  • For version 7.7.20, it appears that some MacOS 10.14 has issues that creates the following error. This error seems to cause crash for the resize and change colors etc. This website (https://github.com/sqweek/dialog/issues/24) basically said this is on Apple's side, so we just have to wait.

  • However, if you use the yml file provided with the package (see the installation instruction) to make a conda environment for PeakPo, you will not have this problem.

    objc[47008]: Class FIFinderSyncExtensionHost is implemented in both
    /System/Library/PrivateFrameworks/FinderKit.framework/Versions/A/FinderKit (0x7fff9f905210) and 
    /System/Library/PrivateFrameworks/FileProvider.framework/OverrideBundles/
    FinderSyncCollaborationFileProviderOverride.bundle/Contents/MacOS/FinderSyncCollaborationFileProviderOverride 
    (0x12db19dc8). 
    One of the two will be used. Which one is undefined.
    
  • For version 7.7.20 or higher, do not try to fix the problem below. Just live with it for now. It is not harmful. If you update h5py, it will break your environment and PeakPo will not work.

    /Users/DanShim/anaconda/envs/peakpo7720/lib/python3.6/site-packages/h5py/__init__.py:34: 
    FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. 
    In future, it will be treated as `np.float64 == np.dtype(float).type`. 
    from ._conv import register_converters as _register_converters
    
  • For version 7.7.20, ignore the warning below. It is because of Mojave, if you use macOS. There is nothing we can do for now. It is not harmful at the moment.

    WARNING:pyFAI.opencl.common:Unable to import pyOpenCl. 
    Please install it from: http://pypi.python.org/pypi/pyopencl
    

2019/02/14

  • For version 7.7.20, we noticed segfault 11 error for three MacOS machines with different OS versions so far when the main window is resized. I have a customized version to get around this problem. It could be related to recent updates as other old installations are fine.

  • Here is a simple fix. Basically we want the window size fixed. Go and open ./peakpo/view/qtd.py. Make the following change in line 18.

    #MainWindow.resize(QtCore.QSize(sizeObject.width()-50, sizeObject.height()-50)) #931, 750
    MainWindow.setFixedSize(QtCore.QSize(sizeObject.width()-50, sizeObject.height()-50)) #931, 750
    #MainWindow.setMinimumSize(QtCore.QSize(10, 750)) #0, 750
    #MainWindow.setMaximumSize(QtCore.QSize(16777215, 16777215))
    

2019/02/14

  • For version 7.7.20, Do not upgrade pyfai and Stick with version 0.15.0, if you have old dpp files. The dpp is version dependent particularly for pyfai. This is a nasty problem. I should address this better in the future.

    pip install pyfai==0.15.0 --user
    

2018/05/13

  • From version 7.7.20, PeakPo can work with matplotlib 2.2.x. But it won't work with matplotlib 2.1.x (not tested). To use matplotlib 2.1.x, you need to stay with PeakPo 7.7.19 or older.

2018/03/07

  • Do not use matplotlib 2.2.0 yet. It is not fully compatible yet. Stay with 2.1.x.
  • Use matplotlib 2.1.x in order to avoid crash when the tweak and ucfit steps are used.
  • Use PyQt5 v5.6.2.

2017/09/16

  • Do not use conda to install pyfai. I had an error for conda install -c cprescher pyfai.
Clone this wiki locally