minidumper - Python crash dumps on Windows
Posted: 2011-09-29 08:04

If you're writing software on Windows, you've likely come across minidumps. They're a great help when your project encounters a crashing scenario, as they record varying levels of information to help you reproduce the problem.

The main product I work on at my day job, a server written in C++, has had minidump functionality since the beginning. We keep PDBs around for our releases, then when customers encounter a crash, we grab the minidump, match it up with the binaries and PDBs, then try to figure out what the scenario was. I think that's fairly standard operating procedure, and it tends to work alright. Release crash dumps are obviously less helpful than debug dumps, but you can still get enough out of them to get started in the right direction. So while one part of my job has that, the other part - the Python part - has had me wishing for it. So I wrote it.

The extension modules I maintain internally for our server's APIs occasionally come crashing down during our test automation. That's fairly alarming at first since the tests just drop out and you don't get much of an indication of why. Was it the extension? The underlying C++ API? Python itself? The unittest logs are all we have to go off of, so then it's a matter of piecing together what was happening at the time, then either manually re-running it from the REPL and/or attaching the Visual Studio debugger to catch the problem.

In comes minidumper. By importing minidumper and enabling it, you can receive crash dump files whenever your Python process goes down. It's there for you.

import minidumper


Now if you do some crazy stuff and cause a crash in your extension code...

    int x = 1;
    int b = x / 0;
}'ll get a crash dump that will tell you exactly what just happened. In my case, I got example_20110929-071529.mdmp. Now if you open that up in Visual Studio, ideally the one that Python was compiled with, you'll get a look into what happened once you hit F5 (or Debug > Start Debugging).

The first thing you'll see is a popup telling you what the problem was and where it occurred, then Visual Studio will show you exactly where in the code the issue lies. As we all know, division by zero is a no-no, and it crashed. If you hit the break button, you can poke around in a ton of information that was gathered from your crashed process. Depending on what value you gave to the type parameter of minidumper.enable(type=...), which defaults to MiniDumpNormal and has a full list of options here, you'll have different amounts of information.

You can walk around the call stack and see what functions were called with what values, and from there you can inspect variables within a function by hovering over them with the cursor. The Debug > Windows menu contains a whole bunch of other pieces of information, including memory, disassembly, value watching, and more.

As far as examples and tests go, I only have some of the basics down, although I plan on bulking those areas up and coming up with more useful and interesting code to prove this extension's worth. I just threw the source up on, but I'm going to wait on getting it on PyPI until I figure out the best way to organize and distribute it.

If you're looking for more info on minidumps, and were helpful sites, as well as the various MSDN documentation.

The following setup steps are what I do to get started, using the CPython default branch, aka, CPython 3.3. Also note that I'm using a debug-built Python, and telling the minidumper extension to do a debug build as it's what I usually use at work, as well as when I'm working on CPython.

  1. hg clone minidumper-dev
  2. C:python-devcpython-mainPCbuildpython_d.exe build --debug install
  3. C:python-devcpython-mainPCbuildpython_d.exe -m tests

Running the tests will build a tester extension, which contains two crashing functions. Right now, the few tests just call the crash functions with different minidumper.enable settings in order to make sure the right dumps are being created in the right places.

Hope it helps.

Note: Until I fix, the crash windows asking you to debug or close the program will stay around until you click something. Ideally I'll be able to add functionality to temporarily disable Windows Error Reporting for the module, as it currently requires manual intervention while running the CPython test suite on Windows, as :code:`minidumper` does.

PyCon 2011 CPython Sprint Newcomers
Posted: 2011-03-16 15:35

Following up two tutorial and summit days, then three days of the conference, the sprints got off to a great start on Sunday evening. I'm back at home now but wanted to put together a summary of the first two days: A lot of great projects got up on stage to pitch their sprint ideas including Brett Cannon speaking for CPython, letting people know where the sprint would be, mentioning the "dev-in-a-box" CDs, and encouraging people to come out and hack. Within 15 minutes of the end of announcements, we had 7 first-time sprinters eager to dive in and get going right away. The new developer guide was instrumental in getting everyone through the initial setup. The plan was to get a Mercurial checkout and as a starting point, as one of the suggested sprint targets was increasing test coverage. By 6:30 on the first day, we were up to 9 people fully up and running, pouring over the coverage results (which were handily pre-generated on the "dev-in-a-box" CD), and diving into code. Here's what everyone worked on:

  • Alicia Arlen started tackling the expansion of string tests and got a patch written and checked in within first day.
  • Scott Wilson noticed some failing urllib tests on his Mac and got to work on fixing them. After that he started on increasing urllib test coverage.
  • Denver Coneybeare mentioned a dbm patch he made a few days before the sprint, then got it reviewed and checked in. He followed that up with test coverage patches to fileinput and _dummy_thread.
  • Jeff Ramnani came up with several documentation and code changes, along with some tracker triage to get a few older issues closed.
  • Michael Henry spent some time on the email package, including some documentation updates and a port of test_email_codecs to Python 3. He's also working on timeit test coverage.
  • Natalia Bidart noticed several test failures after the initial build and test, then wrote up a few patches to make sure her configuration passes all of the tests. She's also working on logging test coverage.
  • Matias Bordese read the dev guide pretty closely and patched a step that didn't jive with his system. He's currently expanding coverage of the dis module.
  • Robbie Clemons started by reviewing a few issues, then took cgitb up to 75% test coverage by starting a test suite for it.
  • Evan Dandrea came up with patches to posixpath, shutil, and tarfile for test coverage and a few bugs.
  • Jonathan Hartley looked into a unittest issue and wrote up a fix plus tests that got checked in pretty quickly. He's also working on coverage.
  • Piotr Kaspyrzyk used a tool he made to find typos in his research work and applied it to the Python documentation, coming up with several patches and many more on the way.
  • Tim Lesher spent time investigating a pydoc issue that was being discussed on the mailing list about named tuples
  • Brandon Craig Rhodes started by running coverage and ended up diving into the order of imports on interpreter startup to fix coverage results before going further with them. He took the new results and is working on copy test coverage.
Here's a picture of some of the group, hard at work: Many thanks to those listed and everyone else who came out to sprint. Hopefully you learned something new and had a fun time contributing -- the effort is definitely appreciated and we look forward to working with you in the future!

FileSystemWatcher on Python 2
Posted: 2011-02-25 19:11

Alright, alright, you guys win. Enough people emailed me to say they would use a 2.x version of watcher from my previous post, so here it is: version 0.2 now supports Python 2. The changes are pretty simple. The "biggest" part of this happened in changeset 96f3f9e4511c, where I handle a few 2 and 3 specific parts split by ifdefs. It's a few sections of handling Unicode/strings/bytes, and then a small change for 2.x to receive the action number as an int rather than a long. I think I did all of this correctly since it works, but that's a poor definition of "correctly" and my Unicode knowledge is definitely lacking. I haven't done a ton of testing on it, but it seems to work alright in my simple test running between 2.7 and 3.1. If you have any issues with it, feel free to submit them or email me.

The five year project: .NET FileSystemWatcher clone for Python
Posted: 2011-02-18 16:51

In the time I've been using Python, no project has started and stopped, and started and stopped again, more than my goal of writing a file system monitor. Sure, it's a small and simple project in the grand scheme of things that could be accomplished over that time, but I like to finish what I start. The idea originally came from my father, also a Python user, suggesting something to work on, likely to help me learn but it'd also help him out. Years ago he wanted a multi-tabbed text editor with tail -f functionality. I think I was reading through a wxPython book at the time and figured, "sure, I can learn this and make that tool." Started it up, had the shell of a simple GUI written, then came time to get the file system updates. I probably got distracted by something, got hooked on something else, then totally forgot about the whole thing. For whatever reason, this happened every few months. About three years ago I tried to rejuvenate the whole thing and found Tim Golden's great "How do I..." page (pretty sure Dad sent this to me before). He has an example, three of them to be precise, covering exactly what I wanted to do: watch a directory for changes using Mark Hammond's pywin32. Awesome. I got something coded up pretty quickly and took the library in a different direction, using it at work to write a Windows service that would monitor our servers and look for crash dumps and email the team. It was super simple and paid off big time, but I kinda just whipped it together and it was poorly designed. Fast forward to a few months ago. I was bored and looking for something fun to work on -- ah, that file system watcher I've been half-assing for years. I thought to myself, "now that I actually know wtf I'm doing, I should do that, and I'm sure my Dad would get a kick out of it." Somewhere in the middle of all of this I was writing C# and used the System.IO.FileSystemWatcher API, which was really nice. I've always wanted the same functionality in Python and liked what they had, so it would be cool to do what they did. A few blogs around the web claimed the Win32 ReadDirectoryChangesW API was behind the scenes of FileSystemWatcher. True or not, it made sense and I was familiar with that from the Tim Golden examples and my watcher service. I've been writing and reading a lot of C code lately so I started hacking. After reading up on a few things, I came up with a much better C equivalent of what I had in that Windows service. It's multi-threaded, uses IO Completion Ports, and seemed to work pretty well. Pass in a directory and a callable, call the start method, then you'll get callbacks for creating files, renaming files, etc. Sweet, we're on the way. After fiddling around with that a bit, I figured it was good enough to build on. I started writing some tests and had simple things like the following working. [code lang="python"] >>> import watcher >>> import os >>> callback = lambda action, path: print(action, path) >>> w = watcher.Watcher(os.getcwd(), callback) >>> w.flags = watcher.FILE_NOTIFY_CHANGE_FILE_NAME >>> w.start() # Then I opened up vim and created a file called "hurf.durf" 1 .hurf.durf.swp 1 hurf.durf 2 .hurf.durf.swp [/code] That was cool and all, but I want to be able to follow one specific file, or files that match a certain pattern. I also want to be able to set callbacks for specific actions. Hmm, FileSystemWatcher can do that. Maybe I'll just build out a clone and see how it works. One of the first things I wanted to figure out was how to emulate the callback attaching and detaching like on Changed events. I needed a container that supplies += and -=, which is none of them. Easy enough, just inherit from one and provide the __iadd__ and __isub__ operators. Before you get outraged: I know that's "unpythonic", but I'm going for a clone here. Filling in the rest was pretty easy. There's a bunch of properties in FileSystemWatcher that map to the attributes and methods of the underlying Watcher. For example, FileSystemWatcher.NotifyFilter sets Watcher.flags, which is an OR'ed group of NotifyFilters, which are constants exposed by watcher from Win32. The weirdest part of the whole thing is that starting and stopping FileSystemWatcher is done by setting EnableRaisingEvents to True or False. It's not a method called start or stop like in the underlying Watcher (or anything else that needs to start and stop). It felt wrong perpetuating this weirdness, and again I know it's "unpythonic", but I'm going for a clone here. As for translating Watcher callbacks into FileSystemWatcher callbacks that work with all of the fancy filtering, it's just a simple queue, a regex, and a big if/elif block. Watcher calls its callback which puts the action and relative path into the queue. FileSystemWatcher pulls it out, sees if it matches the filter, then we figure out from the action which callback to call. If it's a rename, do a special dance, but otherwise create an update object, fill in the details, then start calling back to the user. [code lang="python"] >>> from FileSystemWatcher import FileSystemWatcher, NotifyFilters >>> import os >>> callback = lambda event: print(event.ChangeType, event.Name) >>> fsw = FileSystemWatcher(os.getcwd()) >>> fsw.Created += callback >>> fsw.NotifyFilter = NotifyFilters.FileName >>> fsw.EnableRaisingEvents = True >>> # Opened up Explorer and right clicked to create a new file 1 New Text Document.txt [/code] There you have it. It took 235 lines of pure Python for FileSystemWatcher and 466 lines of C for watcher for this five year project to be completed. If any future employers are reading this, I'm capable of writing more than 140 lines of code per year to complete a five year project, I swear. The project is now on PyPI under the name watcher, complete with a few binary installers. It's 3.x only because 2.x is dead, but I'll do a backport if people are interested (email me: first name at The project is up on bitbucket: It's not really complete but it works pretty well for most usages. I know of a bunch of bugs that I'll eventually fix, but feel free to report more or even fix some of them. Thanks for the idea, Dad.

Why you should go to PyCon
Posted: 2011-01-28 19:55

PyCon 2011 in Atlanta, Georgia, like pretty much all things Python, is awesome. If you've gone before, you already knew this, and hopefully you're joining us again this year. If you haven't gone before, you're about to find out why you should go. Tutorials Starting Wednesday March 9th, the PyCon festivities kick off with two days of two-a-day tutorial sessions, providing you with almost 12 hours of classroom-like interactive education from some of the leading trainers in the biz. Is Django deployment not one of your strong points? Django BDFL Jacob Kaplan-Moss is running a tutorial on it. Interested in stepping your game up with some advanced Python techniques? Raymond Hettinger knows a little bit about that. He's also one of those guys you should just follow around -- you will learn something. Zed Shaw's "Learn Python The Hard Way" will be making an appearance in Atlanta. I've also heard Zed will be available throughout the conference to help you along the way. Python 3 will get some stage time as well through two tutorials Dave Beazley is running. He's doing a repeat of last year's Python 3 I/O tutorial, and Brian Jones will join him for a session about cooking up some Python 3. Tasty. The Conference This is the main thing, the heart, the reason people travel from around the world. Friday March 11th kicks off the three-day conference. Coming off of a record year of talk submissions, there's a great group of talks lined up, and I think there's something for everyone. There's multi-speaker panels like the Python VM talk to give you a "state of the VM" talk about what they are up to, where they are going, etc. There's a talk about optimal aircraft engine tuning. I'm serious, they use Python for that. Is your boss not letting you build out your ideas in Python? Hear from experienced Python users their stories of getting the language into their workplace - from non-profits to schools to big-time megacorps - they've done it. Want to watch a guy do downright diabolical things with a computer from 1979 that's I/O system is an RCA audio jack? Yeah, that'll happen. Python 3 + zeromq + 1979 = "whoa, dude". Also, yay cloud. Speaking of zeromq, Zed Shaw is talking about it. Ya like MongoDB? Got it covered. CouchDB as well. If you don't test, you should. Period. Tox isn't a bad way to do it either. Unit tests are good too. Do you do any of that mobile web stuff kids are into these days? Test it. You ever see those massive telescopes that can see water on Mars or whatever those geeks are up to at NASA? Maciej from the PyPy team does that kinda stuff and he runs it through PyPy. Dave Beazley will not talk about the GIL this time. Almost better than the conference itself -- the hallway track. So, you know we have all of these scheduled talks, and they are great. They really are. However, sometimes you just can't beat standing in the hallway chatting with your fellow Python users. How often do you get to talk to Alex Martelli? Probably not often. How many times a year do you chat with Michael Foord? Not enough. That dude has an awesome beard and he's kinda smart. Get involved in the conversations you see going on -- you'll probably hear about some cool stuff, find out where people are going to dinner, and you'll meet some new contacts. Use your network. Find jobs. Find business partners. Find friends. It's all there. The Sprints After the conference is over, Monday March 14th is when some of the best stuff happens. We didn't all fly to Atlanta with our shiny laptops just to talk about code -- we're also doing some work. Through Thursday, any projects are welcome to hang out and sprint on whatever topic they want. I'll be working on the core sprint like last year. PyPy will probably be there doing some crazy things to make themselves even faster. Most of the web frameworks get together as well. has some information, and as groups announce their presence at the sprints, I'll update this. Feel free to join an existing group or start your own -- the more the merrier. If you are holding a sprint at PyCon, let the PSF Sprints group know at We're still working out how we're going to run this, but drop us a line and we'll keep you in mind. Overall PyCon really is a great time and I've been excited about it for a while now. The tutorials are awesome. The conference is awesome. The sprints are awesome. The people are awesome. The dinners are awesome. It's just a fun time, and if it sounds like a good time to you, now is a good time to buy tickets. Looking to cut costs: check the room sharing wiki. Wondering about transportation? Check out the venue page. If you fly into Hartsfield-Jackson airport, it's like a 30 minute train ride. See you there. ps. I disabled comments because this is an awful WordPress blog. I don't know anything about the internet.
Contents © 2014 Brian Curtin