David Stanek's Digressions - Latest Comments

Re: Measuring Cyclomatic Complexity Of Python Code

Juho Vepsäläinen — Thu, 21 Jan 2010 10:35:37 -0000

Nice work! I just gave it a go in a casual project of mine and it didn't point out any issues. (It's a simple project. :) )

You should add pygenie to PyPI.

Re: Setting Gmail As The Preferred Mail Reader In Gnome

Matt Wilson — Sun, 29 Nov 2009 09:48:48 -0000

Thanks!

Those smartypants at sourceforge ought to make it easier for me to use apt-get to grab .deb files from their projects.

Downloading deb files and installing them manually just seems silly in 2009.

Re: Getting Branch Coverage In Nose

David Stanek — Sun, 15 Nov 2009 17:42:50 -0000

@Ned

That's a great idea! Just sent emails to those lists.

Re: Getting Branch Coverage In Nose

Ned Batchelder — Sun, 15 Nov 2009 11:00:04 -0000

Dave, thanks for doing the work! You should announce this on Testing-In-Python and nose-users...

Re: Getting Branch Coverage In Nose

David Stanek — Sat, 14 Nov 2009 11:02:52 -0000

These changes will not work with older versions of coverage. It would not be hard to fix if there is a need.

Thanks Bill, John and John for bringing that to my attention.

Re: Measuring Cyclomatic Complexity Of Python Code

Matt Doar — Fri, 28 Aug 2009 11:53:37 -0000

Thanks, handy code, worked first time.

Re: Article Help - Dependency Injection in Python

Carlos Ble — Fri, 28 Aug 2009 08:51:08 -0000

DI is fundamental for me as a TDD developer. I can't find a better way to manage dependencies among collaborators. No matter if it is Python, Java or C#. Test driving the software changes it all.
By the way, Pinsor is very nice: http://www.carlosble.com/?p...

Re: Hacking On A super Alternative

Evan Fosmark — Sat, 01 Aug 2009 13:39:14 -0000

Very neat solution. Not saying I'd use it (it would probably confuse the hell out of anyone who saw it unless they knew that super() was overridden), but cool nonetheless. :-)

@Hanno, that's very cool and I hadn't heard about it. One thing that seems poor about it is how you can freely call methods through super() implicitly. For instance, if I have a method foo() and I call super() in it, it'll call the foo() in the supercalss without having to specify super().foo(). The implicit nature of this specific case seems like it could be counter-productive.

Re: Hacking On A super Alternative

Tony Arkles — Sat, 01 Aug 2009 09:43:35 -0000

Err... that was @Hanno, not @dstanek

Re: Hacking On A super Alternative

Tony Arkles — Sat, 01 Aug 2009 09:43:00 -0000

I was mulling over something like this the other day too. Cool to see that I was on the right track (but missing some of the implementation details...)

@dstanek: nice! I was hoping that it was one of the warts that would get fixed, although I haven't actually tried Python 3 yet...

Re: Hacking On A super Alternative

David Stanek — Sat, 01 Aug 2009 07:58:22 -0000

@Joshua: No, it would not break anything.

@Hanno: Very cool. Maybe I should start looking and seeing what other goodness awaits me. So far I have considered 3.x to be in the land of fairies and unicorns.

Re: Hacking On A super Alternative

Hanno Schlichting — Sat, 01 Aug 2009 06:51:49 -0000

How about you look at http://www.python.org/dev/p... ? This exact change has long been implemented in Python 3 ;-)

Re: Hacking On A super Alternative

Joshua Kugler — Sat, 01 Aug 2009 02:43:43 -0000

Knowing absolutely nothing about Python internals: is this simply Python compliant, or will it break on anything besides CPython?

Re: A figleaf Text Coverage Report

David Stanek — Sun, 05 Apr 2009 02:01:47 -0000

That's great news! I am looking forward to the next official release.

Re: A figleaf Text Coverage Report

Titus Brown — Sat, 04 Apr 2009 23:25:40 -0000

Hey David, it'll be in the next release, in some form or other -- I just checked it into figleaf's github repo. Thanks!

cheers,
--titus

Re: I’m Planning a PyCon 2010 Talk

David Stanek — Mon, 30 Mar 2009 12:03:42 -0000

@Graham

Sounds like it yes. I am not familiar with that. This style of virtual hosting is very nice and allows our websites to easily scale.

Re: I’m Planning a PyCon 2010 Talk

Graham Dumpleton — Sun, 29 Mar 2009 17:27:03 -0000

So, effectively doing what the so called virtual host monster in Zope did. That is, handle the virtual hosting within the application itself in same memory space rather than using an external generic mechanism which splits different virtual hosts into different processes.

Re: I’m Planning a PyCon 2010 Talk

David Stanek — Sun, 29 Mar 2009 09:14:21 -0000

@Graham

At AGI we have a single application server that is able to host all of our websites at the same time in the same process. Our new approach is to use a series of WSGI pipelines. I was thinking about talking about that setup.

Re: I’m Planning a PyCon 2010 Talk

Graham Dumpleton — Sun, 29 Mar 2009 06:29:06 -0000

When you say 'WSGI', what particular WSGI hosting mechanism are you talking about? Or is the intent to give an overview of the various options available.

Re: I’m Planning a PyCon 2010 Talk

dave noyes — Sat, 28 Mar 2009 17:01:06 -0000

Very cool Dave. I hope to be at PyCon 2010 and attend your talk. thumbs up

Re: I’m Planning a PyCon 2010 Talk

Alex Martelli — Sat, 28 Mar 2009 14:53:11 -0000

+111 for dependency injection and snake-guice -- and PLEASE accompany that with a concise but powerful coverage of WHY monkey patching is NOT a good alternative if you can avoid it!-)

Re: Article Help - Dependency Injection in Python

Doug Hellmann — Wed, 18 Mar 2009 14:11:03 -0000

Juri Pakaste did an article for Python Magazine that might be useful as background material: http://pymag.phparch.com/c/...

Re: Remap Caps Lock and Left Control on Windows XP

mubed — Wed, 11 Feb 2009 23:05:45 -0000

There is also a way without rebooting, but needs an external tool:
http://vim.wikia.com/wiki/R...

Re: Article Help - Dependency Injection in Python

Greg — Thu, 22 Jan 2009 22:06:22 -0000

I think DI provides the right level of indirection that lets you write simple apps, and then later layer on useful services. As project lead for Spring Python, I have found the ability to apply AOP interceptors to already written apps makes it easy to apply transactions to business logic, put security in place where it wasn't before, and convert my simple app into a multi-node app very easy without altering the method signatures or original business code.

If you have ever coded database transactions, things can get messy. And you have to do it right to handle errors appropriate. With DI, it is easy to plug in a transactional interceptor without the business code having to be aware of it. The only part of your app that has to be in the know is the part that launches your app, because it needs to launch the DI container which holds your blueprints.

The risk of using the term "framework" is that many people interpret it to mean "extend my classes" to get the power. Spring Python provides a DI library that does NOT require you to extend any classes. Basically, it promotes usage of DI principles without the major requirement of hanging your whole application off the framework.

There is no arcane syntax involved. You can pick to create your objects using constructor calls to init, or just setting object properties by name. You may be assuming I'm talking about some XML definition. True, I have one, however, this is python, so why use XML at all? That is why I also wrote a DI container that uses pure python code and a decorator. Create your objects just like you would normally, only now harvest the benefits of have a central place to see how your components are wired together.

Re: Simple Random Number Generator For #codemash

Greg — Thu, 15 Jan 2009 09:30:49 -0000

Clearly whoever organized the #codemash drawing doesn't understand what random.org is or what a random number is.
Random numbers repeat. Each new number is completely random. It could be any possible number. There should be no correlation with any previous sequence.