To get any value from this post you really have to go and read the wonderful article by Venkatesh Rao The Gervais Principle, Or The Office According to “The Office”.

It was possibly one of the best articles written last year, and certainly one of the two best things I have ever read on the disease of organisations (not that this article or the other one were actually about that).

He links his theory to the TV series “The Office”, though it doesn’t have any direct connection to that; it’s just that that is where he saw the clearest examples (another example of satire being truer than reality).

A very basic summary is:

Sociopaths, in their own best interests, knowingly promote over-performing losers into middle-management, groom under-performing losers into sociopaths, and leave the average bare-minimum-effort losers to fend for themselves.

My random notes follow. This article might be edited in the future as I think of more things.

Sociopaths

• naturally distrust group thought/morality
• emphasis on independent thinking
• risk friendly
• manage the big picture

Clueless

• self-deceitful rationalisers
• believe in the group/organisation
• value the opinions and beliefs of others
• risk averse
• manage the people directly below them

Losers

• follow the group/organisation without believing in it
• have accepted their inability to affect change
• are the people who actually produce things

The people reading this will be Losers. Sociopaths have better things to be doing, and the Clueless will react very badly to material such as this. It breaks their suspension of disbelief, and they will have to either attack it, or ignore the bulk of it’s content.

An example of a Clueless reading the article and having to deliberately miss the point is highlighted in the first comment under this post

Organisational Life-cycle

Looking at the classic organisation life-cycle:

• Experimental startup
• Growth into productive startup
• Growth into small business
• Growth into Beaurocracy
• Death (merger, acquisition, bankruptcy, etc)

Start-ups are started by sociopaths (almost necessarily so). As they start to move into a profitable phase they need to hire losers, the only ones who actually produce anything. By definition, if you are producing something on the orders of someone else, that they are being paid for you to make, you are a loser. You are not getting the best value for your time. (This is the point that Marx went all ape-shit over. He didn’t notice the fact that the Losers/Proletariat can’t make things for themselves because they can’t think for themselves what to make. In this way, they are getting the best value for their time, because they can’t do any better. It just wouldn’t be the best value for other people’s time).

As the company grows, the number of Losers grows. The Sociopaths can’t manage them all. The projects are bedded in by now so they hire, or promote, the Clueless to be a safe pair of hands to keep things ticking over.

In order to be able to live with themselves (they are neither producing anything of value nor in a position of power) the Clueless have to deceive themselves that they are important. They convince themselves that what they do is worth while, and that more people like them are needed.

The levels of Clueless swell as the company grows until they are pushing out the Sociopaths at the top. This is when the company becomes a beaurocratic behemoth that the world at large ridicules. This is when you start to see the death throes of the company, and when outsiders start questioning how the company manages to keep going.

Soon after the company hopefully dies. Either through plain old bankruptcy or by being bought, merged etc. Sometimes however, this doesn’t happen. Many public sector institutions can’t die. The Civil Service, the NHS or the BBC can’t go bankrupt, can’t be bought, and can’t really be got rid of. Hence they spend a long time in their death throes. In this case the Clueless act as a parasite in an immortal host.

Hardworking, Lazy, Stupid or Clever

From Wikipedia:

As Chief of the Army High Command, Hammerstein-Equord oversaw the composition of the German manual on military unit command (Truppenführung), dated 17 October 1933. He originated a special classification scheme for his officers:

I divide my officers into four classes; the clever, the lazy, the industrious, and the stupid. Most often two of these qualities come together. The officers who are clever and industrious are fitted for the highest staff appointments. Those who are stupid and lazy make up around 90% of every army in the world, and they can be used for routine work. The man who is clever and lazy however is for the very highest command; he has the temperament and nerves to deal with all situations. But whoever is stupid and industrious is a menace and must be removed immediately!

I think this breakdown of people works well also. Instead of the stupid/clever axis though, I would be tempted to use something more like self-aware/self-decieving. What we traditionally think of as cleverness doesn’t really come into it.

This lead me to create this diagram:

There was a clear gap in the bottom right, that I filled in with the idea of the Sociopath Loser. The Sociopath Loser has the mistrust of organisations of the Sociopath, but without the ability to “play the game”, and general settles for a role outside the organisation, or plays a Loser, writing articles on what is wrong with organisations in their spare time.

This then lead to create this version of the hierarchy:

Thoughts

The Loser Sociopath, such as a drifter, is distinctly a separate class. This is separate again from the unpromoted Sociopaths, and unpromoted Clueless, who are temporarily amongst the Losers.

This ‘organisation’ is also society at large, where the Clueless and Losers make up society, and the Sociopaths, the leaders, are separate from it. They can’t make decisions on behalf of a society that they are part of. Governments aren’t entirely made up of Sociopaths. A great deal of Clueless have made it into reasonably high government positions, without being leaders or gaining power.

The Clueless could be there to provide a cushion between the Sociopaths and the Losers, or it could be that Sociopaths promote the Clueless as a human shield, or that Sociopaths put the Clueless there to keep things ticking over. The Clueless are risk averse and so wont take stupid risks.

I think the rationalising of Clueless is a key thing. They are in a difficult situation. If they were clear minded and honest with themselves, they would see that they are in a bad situation where they have no power or freedom, and yet don’t have the pride of producing anything. Necessarily the only way they can continue what they are doing is to lie to themselves. I’ve seen this myself many times, where people will go to extraordinary lengths to convince themselves why what they have been told to do is the Right Thing. Quite often they have misunderstood what they have to do, but then go about rationalising the wrong thing.

Personal

I am a Sociopath Loser. I have toyed at the edges of the Clueless and ran away in horror. I have also at times done the classic Unpromoted Sociopath thing of putting in less than the minimum effort; preparing myself for the next thing.

I think almost anyone who has commented on any of the articles is a Sociopath Loser, or maybe a Loser. True Sociopaths have better things to do. Clueless wouldn’t be able to do anything but attack or misunderstand. Losers generally wouldn’t have it in them to understand the article.

Further Reading

A Follow up article that Venkatesh wrote

A weird, but intelligent, response that someone wrote. The comments are well worth reading. I loved the opener “While I do speak PoMo well enough to understand most of what you wrote…”

A good response on how the theory applies in religious organisations, and of how it has already been of genuine use to someone

However Emacs is better, cause Steve Yegge says so, and so I switched. I used yasnippet for while, and loved it. It’s a reworking of TextMate’s snippet system with added Lisp; more power without a sacrifice in ease of use. For some reason I stopped using, but since a StackOverflow question recently inspired me, I’m back with a vengence.

A yasnippet to wrap a region or the point in a html tag:

(defun wrap-region-or-point-with-html-tag (start end)
  "Wraps the selected text or the point with a tag"
  (interactive "r")
  (let (string)
    (if mark-active 
        (list (setq string (buffer-substring start end))
          (delete-region start end)))
    (yas/expand-snippet (point)
                        (point)
                        (concat "<${1:p}>" string "$0</${1:$(replace-regexp-in-string \" .*\" \"\" text)}>"))))

(global-set-key (kbd "C-W") 'wrap-region-or-point-with-html-tag)

I wrote this a while, and at the time it was one of the most complicated pieces of Lisp I had written.

I initially misread the StackOverflow question, and submitted the previous function. However once I realised that they wanted something more complex, I felt challenged and dived right in. I eventually came up with this, which hasn’t really been tested:

(defun wrap-lines-in-region-with-html-tag (start end)
  "Wraps the selected text or the point with a tag"
  (interactive "r")
  (let (string)
    (if mark-active 
        (list (setq string (buffer-substring start end))
              (delete-region start end)))
    (yas/expand-snippet
     (replace-regexp-in-string "\\(<$1>\\).*\\'" "<${1:p}>"
      (mapconcat
       (lambda (line) (format "%s" line))
       (mapcar
        (lambda (match) (concat "<$1>" match "</${1:$(replace-regexp-in-string \" .*\" \"\" text)}>"))
        (split-string string "[\r\n]")) "\n") t nil 1) (point) (point))))

From the end of the function working back (which seems to be the way with Functional style), it:

• splits the region by linebreaks
• wraps each line with the yasnippet syntax
• formats each entry as a string
• concatenates all those strings
• replaces the first yasnippet marker with a yasnippet default marker.

I think I should be able to do something better instead of (lambda (line) (format "%s" line)) but I don’t know what.

Definitely the most complex lisp I’ve written to date and I’ve learned quite a lot from writing it.

As always, improvements and suggestions welcome; help yourself to any code.

Quite a few bits of it are copied from the Mac instructions, so it might be a little repetetive if you’ve already read that. (No new jokes, I’m afraid).

Edit: I’ve updated this having tried the installation a few more times. I’m still not 100% sure, but I appear to have got it working several times in a row.

Summary

• Install unixodbc and unixodbc-dev
• Install the patched version of freetds
• Install tdsodbc
• Install the patched version of freetds again
• Configure unixodbc
• Install python
• Install sqlalchemy and pyodbc

Installing freetds on Linux

I’m basing this on a bare bones install of Debian Lenny (ie starting with the bare minimum you can install).

I initially setup sudo, git, svn, bzr, openssh and emacs22. I won’t go too much into them here. You might use different tools. I did place my /etc and /home under version control with git as soon as possible, which I highly recommend for experimental setups. I regret not putting /usr under version control as well.

At some point you will need the tools to compile source such as python2.6 and pyodbc. You might as well install them now.

sudo aptitude install build-essential

I also recommend installing checkinstall which allows you to install from source as if it were a proper debian package. This allows you to meet dependencies with it and remove it easily at a later date.

 sudo aptitude install checkinstall

Next you’ll need to install unixodbc. You’ll also need the unixodbc dev headers for later (eg compiling pyodbc).

sudo aptitude install unixodbc, unixodbc-dev

Then you need to install the patched version of freetds. I pass two options to configure which make it work better with SQL Server, but might stop it working as well with Sybase and others. I don’t really care about that at the moment, but bear it in mind.

mkdir src
cd src/
wget ftp://ftp.ibiblio.org/pub/Linux/ALPHA/freetds/stable/freetds-patched.tgz
tar zxvf freetds-patched.tgz
cd freetds-0.82.1.dev.20090618/
./configure --enable-msdblib  --with-tdsver=8.0
make
sudo checkinstall
# When asked edit the 'provides' (11) value to be 'freetds-common'
# This allows it to be used as a dependency for tdsodbc

Now install tdsodbc:

sudo aptitude install tdsodbc

Then install tdsodbc again:

cd src/freetds-0.82.1.dev.20090618/
sudo make clean
./configure --enable-msdblib  --with-tdsver=8.0
make
sudo checkinstall
# When asked edit the 'provides' (11) value to be 'freetds-common'

I’m not entirely sure about this, but it appears to work for me. It doesn’t entirely make sense though, so I suspect that there is something better that can be done here.

Now you’re ready to setup up your odbc drivers and data sources.

You’ll need to create two files, tds.driver.template and tds.dsn.template. It doesn’t really matter where you create them, as they are templates to create your odbcinst.ini and odbc.ini files, but it’s worth putting them somewhere safe so that you can refer back to them later.

touch tds.driver.template
touch tds.dsn.template

Edit them and add the following, adjusted for your servername, dbname etc.

# tds.driver.template
[TDS]
Description     = FreeTDS Driver for Linux & MSSQL on Win32
Driver          = /usr/lib/odbc/libtdsodbc.so
Setup           = /usr/lib/odbc/libtdsS.so

# tds.dsn.template
[my_dsn]
Description     = Connection to windows virtual machine
Driver          = TDS
Trace           = No
Database        = my_database_name
Server          = MY-SERVER
Port            = 1433
TDS_Version     = 8.0

Note that in the driver template you are naming a driver. You then use that name in the dsn template as the value of the driver. This can be any name you want instead of TDS, but they must match. Alternatively, you can specify the path to the driver in the dsn template.

Now you are ready to install the two templates. You’ll need to run odbcinst for each template. Running odbcinst without any arguments will give you some useful info, but I’ll summarise what you need to know to get it working.

 sudo odbcinst -i -d -f tds.driver.template
 # -i says you want to install
 # -d says you are installing the driver
 # -f says what template you are using

 sudo odbcinst -i -s -l -f tds.dsn.template
 # -i says you want to install
 # -s says you want to install dsn
 # -l says you want to install system dsn
 # -f says what template you are using

These will copy the contents of the templates to /etc/odbcinst.ini and /etc/odbc.ini respectively. They also perform some piece of magic that I haven’t worked out yet (eg. if you duplicate the entry it adds to odbc.ini and change the name slightly, it doesn’t seem to work). However adding a new entry using odbcinst does work.

With the second call to odbcinst, you have a choice between installing system dsn to /etc/odbc.ini using -l or user dsn in ~/.odbc using -h. I recommend system dsn for development, and doing more research for production.

Don’t forget that you might have to add your db-server to your hosts file, depending on your network setup. Just edit /etc/hosts and add the ip address and name of your server underneath the entry for localhost.

Now you should be able to run queries on your database, using isql.

isql my_dsn username password

This should bring up an SQL prompt that allows you to execute queries on the server. Try a couple of queries to test it. If you don’t have a legacy database with data already in it, why the hell are you using SQL Server?

Connecting Python to SQL Server through pyodbc and freetds on Linux

You should now be ready to get python sorted out. If you’ve had any problems so far, let me know and I’ll try to help.

First, you need a python installed. I’ve installed 2.6 from source, which I’ll briefly cover, but you should also be able to use the python2.5 package that’s available.

In order to have all the functionality of python when installing from source, you will need some extra libraries:

sudo aptitude install libncursesw5-dev libreadline5-dev libssl-dev \
libgdbm-dev libbz2-dev libc6-dev libsqlite3-dev libdb-dev tk-dev
# That's all on one line, not two

I’ve used checkinstall to install python as a package. Alternatively you can install it to your home folder by passing --prefix=/home/username to configure.

To install it to your home folder, just download the source from the python website, unzip it, configure it, then use checkinstall.

wget http://www.python.org/ftp/python/2.6.2/Python-2.6.2.tgz
tar zxvf Python-2.6.2.tgz 
cd Python-2.6.2/
./configure
make
sudo checkinstall

I also strongly recommend using virtualenv and virtualenvwrapper. Create a virtualenv as soon as you have installed python and always create your other virtualenvs from that, so that you can easily get back to a clean python if you need to. (I’ll add instructions for that as a seperate post sometime). Install pyodbc into your virtualenv:

    easy_install pyodbc

If you installed the build-essential and unixodbc-dev earlier you should have what you need to compile it.

Now checkout the 0.6 branch of SQLAlchemy. This is currently a branch but should soon move to trunk. It’s fine for development. If you are using this as a guide to setting up your production server, you are insane.

svn checkout http://svn.sqlalchemy.org/sqlalchemy/branches/rel_0_6
# Make sure your virtualenv is activated then:
cd rel_0_6
python setup.py develop

You should now be able to run some basic tests using SQLAlchemy:

>>> import sqlalchemy as sa
>>> uri = "mssql://username:password@my_dsn"
>>> engine = sa.create_engine(uri)
>>> select_query = sa.text("SELECT * FROM MyTable")
>>> result = engine.execute(select_query)
>>> result.fetchall()
[]
>>> insert_query = sa.text(
    "INSERT INTO MyTable (id, text_field) VALUES (:id, :text_field)")
>>> result = engine.execute(insert_query, 
        id="Brian", text_field="Naughty Boy")
>>> result = engine.execute(select_query)
>>> result.fetchall()
[('Brian', 'Naughty Boy')]

That should all work. At this point you should probably try running some other tests. Particularly ones involving non-ascii chars. Then you can watch it crumble and die. (Actually if you’ve installed it correctly, it should work with non-ascii chars).

The UK Pound Sign (£) is a particularly good symbol to use for testing as it has a different byte representation between latin-1 and utf-8.

I have unfortunately had to work with a legacy SQL Server database, and was reluctant to use Windows to do the development. As such I have been trying to get my Mac development machine and a Linux server to connect to the database. It has been quite an adventure, and this is what I think I have worked out so far.

Installing freetds on Mac

I used MacPorts for installing packages. It’s not perfect but it’s a lot easier than installing from source when there is a lot of dependencies, and quite a bit of complex configuration has often been done for you.

Install freetds with the +mssql and +odbc variants.

sudo port install freetds+mssql+odbc

Also install unixODBC

sudo port install unixODBC

You now have a folder /opt/local/etc/freetds/ which should contain two templates tds.driver.template and tds.dsn.template as well as freetds.conf. You shouldn’t need to edit freetds.conf, although you can choose to specify the global tds version and charset (I have done neither of those and it works for me, so best leave it alone unless you have specific problems).

If you don’t have tds.driver.template or tds.dsn.template, create them.

sudo touch tds.driver.template
sudo touch tds.dsn.template

Edit them and add the following, adjusted for your servername, dbname etc.

# tds.driver.template
[TDS]
Description     = FreeTDS Driver for Linux & MSSQL on Win32
Driver          = /opt/local/lib/libtdsodbc.so
Setup           = /opt/local/lib/libtdsS.so

# tds.dsn.template
[my_dsn]
Description     = Connection to windows virtual machine
Driver          = TDS
Trace           = No
Database        = my_database_name
Server          = MY-SERVER
Port            = 1433
TDS_Version     = 8.0

Note that in the driver template you are naming a driver. You then use that name in the dsn template as the value of the driver. This can be any name you want instead of TDS, but they must match. Alternatively, you can specify the path to the driver in the dsn template.

Now you are ready to install the two templates. You’ll need to run odbcinst for each template. Running odbcinst without any arguments will give you some useful info, but I’ll summarise what you need to know to get it working.

 sudo odbcinst -i -d -f tds.driver.template
 # -i says you want to install
 # -d says you are installing the driver
 # -f says what template you are using

 sudo odbcinst -i -s -l -f tds.dsn.template
 # -i says you want to install
 # -s says you want to install dsn
 # -l says you want to install system dsn
 # -f says what template you are using

These will copy the contents of the templates to /Library/ODBC/odbcinst.ini and /Library/ODBC/odbc.ini respectively. They also perform some piece of magic that I haven’t worked out yet (eg. if you duplicate the entry it adds to odbc.ini and change the name slightly, it won’t work). However adding a new entry using odbcinst does work.

With the second call to odbcinst, you have a choice between installing system dsn to /Library/ODBC/odbc.ini using -l or user dsn in ~/.odbc using -h. I recommend system dsn for development, and doing more research for production.

Now you should be able to run queries on your database, using isql.

isql my_dsn username password

This should bring up an SQL prompt that allows you to execute queries on the server. Try a couple of queries to test it. If you don’t have a legacy database with data already in it, why the hell are you using SQL Server?

Connecting Python to SQL Server through freetds on Mac

You should now be ready to get python sorted out. If you’ve had any problems so far, let me know and I’ll try to help.

First, you need a python installed. I’m using python2.6, but 2.5 should work fine, and maybe 2.4. I’m not going to go into installing python in too much depth, but I recommend not using the Mac system python, and also not using MacPorts to install python. I also strongly recommend using virtualenv and virtualenvwrapper. Create a virtualenv as soon as you have installed python and always create your other virtualenvs from that, so that you can easily get back to a clean python if you need to.

Install pyodbc into your virtualenv:

    easy_install pyodbc

Pyodbc needs to be compiled, so you will need a version of GCC. Either install the latest version through MacPorts (this takes a very long time) or install XCode which includes Apple’s own version of GCC. You may also need the python-dev headers and/or the unixodbc-dev headers, depending on how you installed python and unixodbc.

If you have gcc errors and they include the string “-arch i386 -arch ppc” somewhere in them, then python is probably trying to build a universal build but doesn’t have the ppc headers. Edit /lib/python2.x/config/MakeFile and delete every instance of “-arch ppc”. (This is assuming you are running an Intel Mac of course).

Now checkout the 0.6 branch of SQLAlchemy. This is currently a branch but should soon move to trunk. It’s fine for development. If you are using this as a guide to setting up your production server, you are insane.

svn checkout http://svn.sqlalchemy.org/sqlalchemy/branches/rel_0_6
# Make sure your virtualenv is activated then:
cd rel_0_6
python setup.py develop

You should now be able to run some basic tests using SQLAlchemy:

>>> import sqlalchemy as sa
>>> uri = "mssql://username:password@my_dsn"
>>> engine = sa.create_engine(uri)
>>> select_query = sa.text("SELECT * FROM MyTable")
>>> result = engine.execute(select_query)
>>> result.fetchall()
[]
>>> insert_query = sa.text(
    "INSERT INTO MyTable (id, text_field) VALUES (:id, :text_field)")
>>> result = engine.execute(insert_query, 
        id="Brian", text_field="Naughty Boy")
>>> result = engine.execute(select_query)
>>> result.fetchall()
[('Brian', 'Naughty Boy')]

That should all work. At this point you should probably try running some other tests. Particularly ones involving non-ascii chars. Then you can watch it crumble and die.

The UK Pound Sign (£) is a particularly good symbol to use as it has a different byte representation between latin-1 and utf-8.

In iChat options set Show status in menu bar, and in Keychain also set Show status in menu bar.

Now when you press (⌃F8) the iChat status in your menu bar will get the focus. Then you can just press (←) once to highlight the Keychain status, (↓) to highlight Lock Screen and (↩) to select it.

This gives you (⌃F8, ←, ↓, ↩) as a rather convoluted and excessively long keyboard shortcut for locking your computer. It is however very handy when you are working in an office, need to go make coffee and want to save on kittens.

history|awk '{a[$2]++} END{for(i in a){printf "%5d\t%s\n",a[i],i}}'|sort -rn|head

71  git
64  cd
45  sudo
41  ls
39  nosetests
35  svn
31  mate
23  couchdb
22  python
22  lla

71 git
64 cd
41 ls
39 nosetests
35 svn
31 mate
29 port
26 couchdb
22 python
22 lla

Anyway, I tried to install CouchDB on Leopard, and hit a few minor issues which I thought I’d document for others trying to install it.

I installed from Macports (always a good first stop). There appears to be a missing dependency on SeaMonkey, so you need to do:

sudo port install seamonkey
sudo port install couchdb

When you first start couchdb it complians that: CouchDB needs write permission on the data directory: /opt/local/var/lib/couchdb or CouchDB needs write permission on the log directory: /opt/local/var/log/couchdb.

I tried doing sudo couchdb to get over that, and couchdb seemed to start happily, and there was a response from http://localhost:5984/. However, when I inserted my first document using couchdb-python, python hung completely with no repsonse (I left it for 10 minutes while I made a coffee). In retrospect, sudo couchdb probably wasn’t the correct way to go (particularly as those directories don’t even exist), but I was keen to start playing. The correct thing to do is probably to create a couchdb user that runs the database and do sudo -u couchdb couchdb to start the database. Unfortunately I’m quite lazy, so instead I just did:

sudo mkdir -p /opt/local/var/lib/couchdb
sudo chown ed /opt/local/var/lib/couchdb
sudo mkdir -p /opt/local/var/log/couchdb
sudo chown ed /opt/local/var/log/couchdb

Where ed should be replaced by your username. mkdir -p recursively makes directories (incase you don’t have /opt/local/var/lib/ or /opt/local/var/log) yet.

Anyway, hope this helps someone else, if they have problems.

It’s based on Django and Paste and seems to be inspired by bits of web.py and maybe bits of Werkzeug.

The templating is directly Django’s, which I think is a shame as Jinja is very similar but better (vaguely useful error messages for a start).

The model is inspired by Django, but using Google’s Datastore (not an rdb), using GQL, their modified version of SQL. It wasn’t easy to get going with, which is partly the fault of the documentation (it took me ages how to work out how to do their version of a primary key).

Once you get the hang, it’s pretty simple for quick simple data. It would be tough going to work out how to do complex relationships with it, particularly for anyone from a traditional rdb background (like me). There’s no section where they explain how their datastore works and the concepts behind it. I’ve been playing with Couchdb recently and they did a wonderful job of explaining the overall concept before going into how to use it. That kind of conceptual overview for Datastore would probably make a world of difference.

The controllers seem to be inspired by Web.py. They are classes with methods for POST, GET, etc. I like that system from playing around with Web.py. It allows you to very easily use the same URL for displaying an edit form and saving an item. It keeps your code nicely grouped, but I think it would make it a little more difficult to create generic CRUD controllers (maybe having to use meta-class magic).

The routing seems based on a corrected version of Web.py’s. It fixes the annoying list of tuples into a proper list of lists of tuples. It uses basic regexs for routing (first one that matches wins), and a direct reference to a class. This means the routing has to go at the end of the module (which is different to most other frameworks). I like it and it’s exactly how I set up the routing when I was messing around with WSGI.

Sessions and deployment are the killer features. Deployment is just a single command that uploads all your files, and restarts your app. Session handling uses the Google login system, so sessions are completely abstracted away. These are often the hardest bits to get right when using a new framework, so they’ve done impressively well to make them so simple.

Overall, I’d say it could be outstanding for quickly knocking out an app you’ve just thought of. A simple app will scale beautifully if it suddenly becomes popular (no more slashdotting) and they’ve solved the more annoying problems for webapps. I would have some concerns about trying to create a huge complex app using it, but we’ll see. Certainly I’ll use it again if I ever have a good, simple idea. For my bad, complex ideas I’ll continue building my own framework

As an update, it appears you can use almost any Python modules in your code as long as you add them into your project folder (either copy the package or if you use a symlink it will copy them automatically). That would mean that everything hinges on their storage. Sessions and deployment are excellent, and everything except for the model you can do yourself. If their storage is easy to use with potential longterm power, then it could big, and a serious choice for hosting major sites.

The only other issue would be one of branding. As the login makes it clear that you are using a Google AppEngine, I don’t see big companies using it. That’s a good thing as it gives us open minded charities a leg up on the over-resourced private sector.

WSGI seems excellent and could be great for Python. Frameworks tend to make things seem more difficult and mysterious than they really are. What has taken me a long time to learn is that WSGI is really very simple.

At it’s simplest there are three aspects to it:

• You receive a dictionary with useful information in it (environ)
• You receive a function to call that starts the response (start_response)
• You return an iterator (ie a list) of strings (this is your page body)

Anything on top of that is just helpers. Helpers are great, but you don’t need a framework that hides what is really going on, when it’s as simple as that.

You can create a hello world app as simple as:

def app(environ, start_response):
    start_response('200 OK', [('content-type', 'text/html')])
    return ["Hello World"]

Obviously that just returns Hello World with no html. Which is bad. So:

def app(environ, start_response):
    start_response('200 OK', [('content-type', 'text/html')])
    begin_body = "<html><body>"
    body = "Hello World"
    end_body = "</body></html>"
    return [begin_body, body, end_body]

We don’t want to write begin_body and end_body every time, so lets write a function to handle this for us:

def put_in_body(args):
    begin_body = "<html><body>"
    end_body = "</body></html>"
    return [begin_body] + list(args) + [end_body]

Then we can just:

def app(environ, start_response):
    start_response('200 OK', [('content-type', 'text/html')])
    return put_in_body("Hello World")

We’re going to want to be able to do lots of different things, and we don’t want to write them all in the body of one function, so let’s write a dispatcher:

def dispatch(environ):
    # We need to get the path from the environ, and work out where to go.
    path_info = environ.get('PATH_INFO', '')
    if path_info == "/goodbye":
        return goodbye(environ)
    else:
        return hello_world(environ)


def hello_world(environ):
    return ["Hello World"]

def goodbye(environ):
    return ["Goodbye World"]

def app(environ, start_response):
    start_response('200 OK', [('content-type', 'text/html')])
    return put_in_body(dispatch(environ))

Now if you go to /goodbye it will say “Goodbye World”, otherwise it will say “Hello World”.

As you can see it is all very straightforward. It allows you to do things however you want. Adding more and more functionality is just a case of adding functions that returns strings, and telling the dispatcher when to call that function.

In a later post I’ll talk about threading. It turns out that’s easy as well.

If you want to run the code above, just easy_install paste and then add this to the bottom of your file (which I called wsgi_sample.py):

if name == 'main':
    from paste import httpserver
    httpserver.serve(app, host='127.0.0.1', port='8080')

Then you can do python wsgi_sample.py and visit http://localhost:8080 to see it working (don’t forget to visit http://localhost:8080/goodbye before you leave).