I installed SublimeCodeIntel package in sublime text, and I use it to code in python 3. However, the problem is that it uses python2 paths to do the import, so if a library is installed in python2 and not in python3, i won't find it when i use import or from X import y.
I did confirm that when I used the command SublimeCodeIntel: Dump Import Directories , so I saw both the files python and python3 in the directory ./codeintel/import_dir_stats and here is there content :
python
dedeeb56f744e507026fef17243da41f /home/bilal/.local/lib/python2.7/site-packages
6a1d0cac3d9e6148e2208b63a33a1e6f /home/bilal/.local/lib/python2.7/site-packages/impacket
16a4fccbb3beadfdfd72691ef8f7298c /home/bilal/.local/lib/python2.7/site-packages/mechanize
211d2b55059f6b634799fdae534decd9 /usr/lib/python2.7/dist-packages
be5448890caffe81686310f127d6efae /usr/lib/python2.7/dist-packages/_markerlib
cec69a0830a725e10ac4e364d44add8f /usr/lib/python2.7/dist-packages/appindicator
python3
dedeeb56f744e507026fef17243da41f /home/bilal/.local/lib/python2.7/site-packages
6a1d0cac3d9e6148e2208b63a33a1e6f /home/bilal/.local/lib/python2.7/site-packages/impacket
16a4fccbb3beadfdfd72691ef8f7298c /home/bilal/.local/lib/python2.7/site-packages/mechanize
211d2b55059f6b634799fdae534decd9 /usr/lib/python2.7/dist-packages
be5448890caffe81686310f127d6efae /usr/lib/python2.7/dist-packages/_markerlib
cec69a0830a725e10ac4e364d44add8f /usr/lib/python2.7/dist-packages/appindicator
I didn't put all the lines (because there is a lot), but the content of the two files is identical.
I don't understand from where is coming this problem, why SublimeCodeIntel is looking in python2 directories for the import ??!!
PS : I am using Ubuntu 15.10 (with Linux version > 4), and sublime text 3 build 3103.
Please help, I really consumed a lot of time and energy looking for this.
I would strongly recommend using the Anaconda plugin (no relation whatsoever to the Anaconda Python distribution) instead of SublimeCodeIntel. I struggled with SCI for a while on various machines, and could never get very good code completion out of it until I ran across Anaconda one day. The next day I removed SCI and have been completely satisfied ever since. It's super-easy to configure (just give it the path to your python executable and it reads sys.path and all the rest), and pretty much just works. It has linting with several different linters built-in (you can disable them if you want), has popups available for function signatures and other hints, works with virtualenvs right out of the box, and more.
(I didn't write it and I have no connection to the author(s) - it's just a great plugin!)