Not only did my bank have a similar silly requirement (and the username was just numbers to boot) but they also didn't feel like even having an RSA token for online banking.
The bank I was with before had sane defaults but alas a takeover made me a customer of this new bank.

They told me I shouldn't worry, I told them I followed their stock the last two years something they might concider doing as some sort of feedback mechanism and switched.

a fun one from yesterday.

i'm working on some code which makes frequent use of this pattern:

(btw "with patch.object(...)" is a nifty construction.)

so i come along to add a feature. i write a test making use of this pattern and run it, expecting it to fail. once it does, i'll go fix cluster.activate_node() so that it calls activate() on the cluster's Node object. tdd bitches!

except the test doesn't fail. that's weird. i rewrite the assertion like this:

and this time it does fail. wat.

so i ask the guy who wrote this code if he knows what's going on. shouldn't these two assertions be identical? he agrees they should. we look at a few things before i discover that i cannot find a definition of assert_called_once(). the mock library provides an assert_called_once_with(), but no assert_called_once(). wtf? how is this working?

well, since m_activate is a mock, it has no problem at all executing m_activate.assert_called_once -- the method is simply mocked out! the original author never noticed this because he wrote all his tests post-hoc and never verified that they could fail. he had other assertions that pinned down most of the behavior, so the code did what it was supposed to, but essentially he was lucky.

and that's why you should quit your bitching and learn tdd. at the very least, you need to make sure your tests are able to fail. we'll see if this lesson landed on my colleague.

once discovered, of course, the problem was easy to fix:

(i could have avoided some refactoring from foo.assert_called_once() to assert_called_once(foo) by adding my method directly to the mock library, but that felt a little too magical and hard to read.)

In related news I started TDDing my Prolog code last week. I was mostly superprototyping before but from here on out it'll all be TDD.

Surprisingly little info on Prolog-TDD outside of PLUnit docs that are not exactly example heavy heh, maybe a decent topic for a blogpost for that shiny new blog of mine that is at the status...wordpress installed and semiconfigured but redirection into nowhereland untill actually write the first post

tc,

good post. I've been bitten by a similar bug/test combo...not using mocks, but a test that I wrote after the fact that wouldn't fail no matter what I did

Not really

Barack Obama is doing an AMA on Reddit

http://www.reddit.com/r/IAmA/comment...united_states/

If you're going to post a photo why not at least show some of the screen so you can prove you're actually logged into Reddit?

It's definitely him: https://allthingsd.com/20120829/how-...nsylvania-ave/

I just thought it weird that they didn't show that he was on the site. I'm sure the photo's him and that's a photo from him in the campaign office and all that.

yeah, would have probably made more sense.

Browsing stuff on github and I come across this file, called "core"

Am I missing out for not following Reddit at all?

Probably a little, but it's not worth the time suck. Plus, if you're not some 22 year old raging liberal you'll probably not get that involved.

It can be what you make of it. Most of the default sub-reddits are good for quick hit time kills. However, other have detailed discussions and stimulating content. Some of the defaults also have a high noise to signal ratio, but you can unsubscribe from them. But just like how 2p2's sub forums have grown, there are places where you can go to talk about anything. And you can not go to the sections that tend to be filled with people that annoy you.

Calculate what the chart will display, *then* plot it.

fakesky, i think you want: http://en.wikipedia.org/wiki/Bubble_chart

There's just not enough hours in the day imo. HN provides me with more than enough reading material.

That said, I do glance at the Django & Python subreddits once a week but don't really ever read the commentary like I do on HN.

here's a git error i encountered today:

...

I hate it when I get a bum demultiplexer, all you can do is reload the flangematrixator and hope for the best.

what the **** is a samoflange?

also kerowo's response is basically the same as the first round of responses in irc when this problem was reported.

Just found a page on my intranet app that was generating > 2500 queries to retrieve ~50 numbers

lol ORM.

I recently needed a web app spell checker with the feel of a desktop word processor. I couldn't find one I liked so I wrote one. It's on github, so if anyone needs a spell checker for a web app, check it out. Here's a live demo.

Nice work. The one thing I noticed is that you get false positives for hyphenated words (e.g. ex-wife) but it looks like that should be easy to fix by just changing your regex to allow hyphens as well as a-z.

Fixed