Mark's Blog

To admit it, I was inclined to title this post “from Europe/Russia with love”, but technology should not be seen national and nginx, Varnish and MonetDB are few of many products I am going to mention.

Here’s “nginx” pronounced like you would say “Jeanny”: “Jeannix”.

Let me turn tendencies into trends and provide me you with what I call the weather forecast of technology: Read more »

Tags: apache, GNU/Linux, monetdb, MySQL, nginx, python, server

Once gone programming in Python, perhaps with the impressive TurboGears or Django, developing RESTful web applications with Java might seem like cleaning the bath with a toothbrush at first. It is the switch from a higher language to one a bit less, well, ‘high’. For small projects it might be definitely overkill.

Even Google and NASA (see video with comparison) prefer Python. Why would you want to switch?

In case you must, I’ve collected some equivalencies for you as kickstart. Read more »

Tags: java, python

On implementing double opt-in and -out options for mailing lists I needed “magic tokens”, i.e., strings which are unique for every email address in our databases. Widely used are MD5 hashes of some formatted time strings, such as the current date with microseconds. Or the latter are used to seed an random number generator for generating the MD5 or SHA1 hash.

As the result is a long integer it is classically displayed as hexadecimal. And stored without any further conversion as such – which is IMHO a waste of space. Why not using every 36 letters and digits?

In Python you can generate an unique ID by:

import uuid
uuid.uuid1()
# UUID('3208c170-743b-11dd-a60f-000e354e9618')
uuid.uuid4()
# UUID('cb7b64ca-068f-4590-9886-cf375d26f796')

Converting every part from base16 (hexadecimal) to base10 (decimal) is simple:

int('ff', 16)
# 255

Luckily Aloysio Figueiredo and Kip Bryan have published an one-liner to convert from base10 to any other radix:

def baseN(num,b):
  return ((num == 0) and  "0" ) or ( baseN(num // b, b).lstrip("0") + "0123456789abcdefghijklmnopqrstuvwxyz"[num % b])

Putting it together I got base36 unique IDs for my tokens by this Python code:

import uuid

def baseN(num,b):
  return ((num == 0) and  "0" ) or ( baseN(num // b, b).lstrip("0") + "0123456789abcdefghijklmnopqrstuvwxyz"[num % b])

def uuid1_base36():
    '-'.join([baseN(int(p, 16), 36) for p in str(uuid.uuid1()).split('-')])

Do you see any benefit from displaying and storing hashes in hexadecimal?

Tags: know-how, python, UUID