PyCharm 2017.3 is out now!

PyCharm 2017.3 Released

PyCharm 2017.3 is here: it’s faster, more usable, and better for data science.

Download now

  • PyCharm 2017.3 brings many performance improvements: Python indexing is twice as fast, JavaScript indexing is up to 40% faster, configuring SSH interpreters is much faster (in some cases over 100x), and more!
  • We’ve worked hard to make creating and selecting Python interpreters easier. For example, PyCharm 2017.3 will remember if you prefer to put your virtualenv in your project folder, or in a separate folder (like your WORKON_HOME folder).
  • A new scientific mode, which puts all the tools you need for data science at your fingertips (Scientific mode is only available in PyCharm Professional Edition)
  • We’ve built an all new HTTP client for those of you who build (REST) APIs. You can now write a request in a .http file, and then run it straight from the editor (Web development features are only available in PyCharm Professional Edition)
  • Django 2.0 will be out soon, and we’re happy to announce that PyCharm 2017.3 fully supports the new version (Django support is only available in PyCharm Professional Edition)
  • PyCharm 2017.3 supports running Python modules (python -m <module name>)
  • We’ve added support for SSH config files: if you’re using a complex setup with jump hosts, you can now connect to your machines effortlessly using the same config that you use on the command line (SSH is only supported in PyCharm Professional Edition)

And much more, have a look at our What’s New page, or get PyCharm 2017.3 now and try it yourself.

This entry was posted in Release Announcements. Bookmark the permalink.

34 Responses to PyCharm 2017.3 is out now!

  1. James Kerns says:

    Perhaps the best release by far given these new tools!

  2. avshyz says:

    Amazing work guys!

  3. Natas says:

    How did you loose support for closing a file via [command w]. Don’t you QA your product

  4. yegle says:

    Just FYI, the App Engine support in PyCharm is confusing. It doesn’t work with gcloud SDK (the official Google Cloud SDK), only works with the legacy App Engine SDK.

  5. Kubas says:

    Is “Go to definition” fixed ? In version from 10.2017 or 11.2017 it doesnt work for me and my 2 peers. I have been downgraded to 2017.1.4 beacouse of that (June)

  6. divenex says:

    As a scientist, and heavy Python user, I love the new scientific mode. Thank you very much!

  7. David says:

    Is there an easy way to close all the plots shown in sciview? If I rerun a script that generates a bunch of graphs, it will quickly eat up memory. The only way I see to close plots is by clicking the x next to each, which gets very tedious very fast.

  8. jim says:

    after updating tensorflow brakes.

    ModuleNotFoundError: No module named ‘multiprocessing.pool’

    anybody having a similar issue?

  9. janscas says:

    I have problems with the out of memory issue… Never happend before and I have been using pycharm in the same computer for the last 3 yearsñ.

    please fix this issue.

  10. janscas says:

    WTF… pycharm eating 99% of my CPU, 99% of the sisk usage and upt o 4GB of RAM…

  11. nurgasemetey says:


    Traceback (most recent call last):
    File "/usr/lib/python2.7/encodings/__init__.py", line 74, in search_function
    entry = _cache.get(encoding, _unknown)
    AttributeError: 'NoneType' object has no attribute 'get'

    I am getting this error after update.

    • Paulius says:

      Me too

    • Stefaan says:

      Same here :(

      • Stefaan says:

        I managed to fix it on my setup. The full concrete fix is probably too specific for my use case, but maybe some pointers might help others:

        PyCharm adds /Applications/PyCharm.app/Contents/helpers/pycharm_matplotlib_backend to PYTHONPATH
        this folder has a sitecustomize.py that does “import matplotlib”. This import apparently failed, which has side effect to break the encoding stuff in the stack trace above (I think, not completely sure about that). The reason of the failing matplotlib import was another path X in my PYTHONPATH that provided a matplotlib that was not compatible with the python interpreter I was using. That X path was added because I had buildout support enabled in project settings, using the paths from a script that was not completely up to date with the project. Removing path X from this buildout script fixed the problem for me.

    • Ernst Haagsman says:

      Thank you very much for reporting this issue. We’ve created a ticket, and we’re looking into it. To receive updates on the progress, please vote for this ticket: https://youtrack.jetbrains.com/issue/PY-27481

  12. Ryan says:

    Hi guys,

    Exciting features! One minor issue, though. I breakpoint and open a debug prompt pretty much by habit whenever I debug. There was an “Open Debug Prompt” option that went away with this release. I see a “Show Python Prompt” button (grey, with a >_ symbol) in the debug console window which does it, but I do not see they ability to assign a hotkey for it in the keymap. Is this possible?

  13. Jeks says:

    Will you support pipenv in feature ?

  14. Yotam Michael says:

    After installing the new version standard output is not displayed at the console.
    for example when I type: 1+1 (enter), I just get a blank line.
    as a workaround I use print all the times, anyone have better suggestion?

    • Ernst Haagsman says:

      Thank you for letting us know! This is related to older versions of IPython, we’re already working on resolving this bug: https://youtrack.jetbrains.com/issue/PY-26479

      To get your python console to work again, upgrade your IPython to a newer version. You can do this by going to Settings | Project Interpreter, selecting ipython, and then clicking clicking the blue upward-facing arrow (near the ‘+’ button for installing packages). Alternatively, you can run pip install --upgrade ipython on the command line (make sure that you have the virtualenv activated, if you’re using one).

  15. Gaëtan de Menten says:

    There are **a lot** of great features in this release (some of them not mentioned in this blog post). **Congratulations to the team!**

    Speaking of Python indexing, you touched my second biggest gripe with PyCharm (*). It is a lot better than in previous versions but the problem I have had for ages (driving me nuts at times) is not entirely fixed either:

    1) you cannot run a script while PyCharm is indexing files (and I do not see any reason why that is since PyCharm is not supposed to be involved during the script run).
    2) it is often re-indexing (**) stuff that has already been done. I do not think my setup is that peculiar but I have dozens of conda environments and dozens of projects, 2 to 4 being open at the same time. But when I update an environment that is used by several projects (namely my dev environment which is used by 99% of my projects during development), all projects using that environment do **some** reindexing when I reopen them (I don’t think it is a full reindex because it goes faster) but combined with 1) this is still a pain (though **much** better than in earlier versions).

    (*) the biggest one is the continued lack of partial git commit release after release (https://youtrack.jetbrains.com/issue/IDEA-63201)
    (**) I mean reindexing in a broad sense: any combination of “Indexing”, “updating python path”, “querying skeleton generator”, “scanning installed packages”, “Scanning files to index”, …

  16. Thomas15v says:

    Really love the faster indexing :). However I am still bothered by this issue: https://youtrack.jetbrains.com/issue/PY-21362. Any chance this could be fixed in the next update? It’s already reported a year ago.

  17. Dylan Cissou says:

    while debugging, the dataview panel is not attached to the current debugging session, means none of the data object can be seen. I have to manually click on the object from the variables window, then I could type variable name directly into the dataview.

    • Ernst Haagsman says:

      The data view is linked by default to the Python console. And as you could have multiple concurrent debugging sessions, there’s no way for us to connect to them without more information. Clicking in the debugging pane is the easiest way to inspect a variable.

  18. Dan Drummond says:

    Just upgraded from 2017.2 to 2017.3. Unable to get 2017.2 programs to load from out of memory errors. I have 12 gb of memory on my computer. The last program I tried to load was only 48 lines long and I got an out of memory error.

    It seams that now all programming must be done in a virtual env. If this is correct I REALLY DON’T LIKE THAT !!! It’s had to tell for sure as I can’t seam to get ANY of my python stuff to run from pycharm.

    If this isn’t fix very shortly you can cancel my license!!!

    • Ernst Haagsman says:

      Out of memory for a 48 line program sounds like a bug. Although it depends on what you have installed in your interpreter, are you using Anaconda by any chance? If you upgraded PyCharm using the built-in updater, you could try to reinstall using the full installer. If the problem persists, please send your idea.log file to pycharm-support@jetbrains.com.

      Regarding virtualenvs, this is not necessary. If you really want to develop a program with your system interpreter, you can. Just go to Settings | Project Interpreter, choose ‘Add Local’ to configure an interpreter for this project, and choose ‘System Interpreter’ on the left.

Leave a Reply

Your email address will not be published. Required fields are marked *