PyCharm - first impressions

This is a post copied from my blog at http://weholt.blogspot.com/2011/05/pycharm-first-impressions.html, reposted here at the request of the lead developer of PyCharm. Thanks to yole for his interest and wanting to answer directly to my "review". Mind you, this review is based on very early impressions and I've had little time to use PyCharm. Here we go:

I've been using PyCharm for my django development after I recieved a free open source licence for it based on my work on DSE. I've used free software based editors like Editra so far, but I wanted a visual debugger and the option of running syncdb, the django devserver, and unittests from inside an IDE, spending less time in the shell, more time in the editor. That's what drove me to PyCharm. I've also used several of JetBrains products in my daytime job, like ReSharper, dotTrace and TeamCity, all very good products for C#/.NET development.

A very brief report on my experience with PyCharm so far:

• It feels like a big IDE in many ways; It takes some time to load. It feels heavy. It scans your source code to provide intellisense, but it should be able to cache stuff. When started it looks as if it scans all your python code, even the standard lib and all installed modules all over. Sometimes the intellisense works great, other times it's providing you with some weird match for a module installed by Ubuntu for instance, far far into your lib-folder, not even remotely related to what you're working on. I did like that when I wrote HttpResponse in my views.py and didn't have the correct import statement in my file it imported the correct module from Django when I pressed Alt+Enter, like shown in the image above. Very ReSharper-like.
• It must be configured to be useful and there are lots configuration options. Some of the options makes more sense than other, for instance the "Intentions" section makes suggestions for improvements for your code and it wanted me to change list comprehensions into for loops. Why? Isn't list comprehensions a better choices than loops?
• Getting PyCharm to work with existing Mercurial repositories was hard. This was pointed out by Guido in his review of PyCharm as well. I like that I can run the devserver, syncdb etc inside the IDE, but I didn't get the visual debugger to work and that's one of my main goals. I'll spend some more time looking into this. It would be just great to add breakpoints and debug my source like I'm used to in Visual Studio. Using pdb/set_trace() is doable, but it feels old.
• In the source tree for my project there's a tree item called "External libraries" containing all the available modules, both standard lib and third party. It makes it very easy to dive into the source of a module. Very handy indeed. Some of the refactor methods just doesn't do anything. If it doesn't make sense to refactor something, disable the refactor menu item. Other times the refactoring works like expected.  Intellisense for methods defined in your current source file works fine, auto-completion is snappy for the most part and hints about parameters works fine as well.
• I tried to let PyCharm optimize my imports. Removing unused imports are great, but sometimes PyCharm suggested to remove imports needed because of monkey patching. Still, it work ok in most cases.
• "Inspect code..." gives you a list of errors found in your code. It found unused variables, unresolved references etc. Handy, but I missed having something like PEP8-checker directly integrated. Is it in there? I didn't see anything.
• Tag-completion worked fine in html and there are support for django templates, tags etc, but I didn't test this very much. Based on information here it looks great.
• It also provided easy to use support for unittesting, like shown here, but when I tried to debug my django unittests it failed because of some problems with the configuration. It's clearly my fault, but like I said earlier there are lots of configuration options and I probably messed up somewhere.

All in all I found lots of the stuff in PyCharm to be nice and familiar having a background in IDEs like Visual Studio, but it's also very clear that you have to learn a lot of keyboard shortcuts to be enjoy all that PyCharm has to offer. The thing I like best about programming in python is that in most cases I manage to keep my projects small, easy to navigate and also easy to debug. Working with bigger projects probably justifies buying a piece of software like PyCharm, where the need for code navigation, refactoring etc makes more sense. Getting the visual debugger to work is still high on my wishlist so I'll spend more time looking at that.