Profile cover photo
Profile photo
Robert Lehmann
Robert's posts

Comments on a CVV generator:
«i am a junior credit card thief. and would like to say you guys have made my job a whole lot easier […]  any other criminals who would like to exchange thoughts on matters of fraud .feel free to contact me , !!!»

This sounds like such a bad and obvious FBI sting.

How legitimate expectations trump ‹ignorance is no excuse.›

(I don't have an opinon on Roman Polanski's case, but the Swiss reasoning strikes me as ill.)

In their extradition rejection statement, the Swiss federal authorities wrote:

Roman Polanski would not have decided to come to the Zurich Film Festival in September 2009 if he had not been relying upon the fact that this trip did not legally disadvantage him.

So, with the same reasoning, robbers can now freely attack banks because, hey, they wouldn't do it if they'd expect legal penalties, right?

German politics at full force.

So, there is this license fee for general public access to radio and television in Germany (Rundfunkbeitrag, Broadcasting Contribution Fee.)  Every citizen pays a tax-like monthly rate to fund public media.
Then there is this committee (KEF) which regularly has to revisit the financial requirements of said media in order to recommend adjusting the fee.  The members are appointed by the federal state premier.

The KEF recently, based on a surplus of €1.1 billion, suggested the fee be reduced by 74 cents.  This sounds like a genuine assessment but they stated it still leaves half(!) of the net income to make up for fluctuations (or, uncertainty of the data.)
The very premiers who put those experts in charge to hear a reliable report now ignored that recommendation and instead only reduced the fee by 48 cents.  (This leaves the fee at a convenient monthly €17.50.)  This leaves an excess surplus of about €700 million in the pockets of the public broadcasting companies, and out of the taxpayers' pocket.

I just had a look at the list of reserved domain names for a couple of gTLDs and am confused why there's so many SHA1 hashes of popular domains in there.

See for .schule and for .berlin.

I really wonder how the White House works.  It must go something like this:
“Uh, we need to do something about Russia granting Snowden asylum. Ideas?”
“Nuclear threat?” — “Hell no.”
“Trade embargo?” — “Nope.”
“UN resolution?” — “Meh.”
“Olympic boycott?” — “Sure, whatever.”

I want to take a shower real bad now: I just implemented time travel.
(Disclaimer: This was the thing that I initially thought could be made to work most easily and somewhat wanted to finish as a case in point. There are so many wrongs in this story it gives me shivers as I write this down and I'm deliberately not going to post any code so that nobody ever even thinks about duplicating this.)

We have a couple of Django models which have some kind of a countdown and, in order to achieve that, need the current time: For our tests I wanted to manipulate the current time — a feat commonly known as time travel.

Easy enough, I thought, let's subclass datetime.datetime and monkeypatch it into the datetime module, overriding the now() classmethod as we go.

A mesmerizing combination of circumstances made this approach break in subtle ways: Django, when serializing datetimes, checks whether something is an instance of datetime.datetime — which we purposefully replaced with our mock class. Other view code, in turn, adds datetimes with timedeltas which always returns a classical datetime object no matter what the initial type was.
To put it explicitly: We have original datetime objects flying around, trying to be serialized, but Django does not detect them as datetime objects any longer because:
isinstance(a_real_datetime_object, datetime.datetime) == isinstance(a_real_datetime_object, MyDatetimeMock) == False

And this is where we strike our masterpiece, which is not for the faint of heart: In a moment of mischief and mental derangement, I introduced a metaclass to the field which readily overrides instance checks (_instancecheck_) of the mock class to delegate to the original datetime type. This then makes the instance checks succeed even with the monkeypatched mock class and probably kills a kitten every time it's run. Or is going to have been run, to speak in time travel terms.

I recently heard an interesting discussion on fiction vs. non-fiction literature:

Non-fiction is usually read to learn something practical, like fishing.
Fiction has the opportunity to cover a larger range. It can deliver entertainment but also demonstrate wholly new trains of thought. (Science fiction usually claims this.)
They can transform your entire mental landscape and teach you something very useful -- much like non-fiction. Stories are just illustrative devices in this category of books.

But when there is already enough trouble processing the information of the real world, why would an author ever obfuscate his thoughts in a fictional setup?
Well, often there are enough real life situations for any possible lesson for sure but they often fail to deliver a point so subtle and complicated that it takes an easier device to get through -- parables come to mind.

Now the opposing party argued that stories just blur the actual point being communicated.From here the discussion quickly revolved around the utility of story-telling in general: is there an advantage of wrapping a viewpoint in a made-up story and fake examples to help remembering it or should readers just be presented with a bullet list of lessons learned?

(I'm sorry for the verbosity of this post but I found the points made in the discussion, between +Daniel Tenner and +David Welton in #startups, extremely interesting and wanted to share them.)
Wait while more posts are being loaded