Run To The Future

We have a "master" repo of git (so yes, we're using it in a similar way to cvs or svn) and we keep running into problems, namely when some tries to push to files someone else has pushed:

error: insufficient permission for adding an object to repository database ./objects

We are running git over ssh, which should be the method of choice behind a firewall.

After some investigation I re-initialised the remote repo with:

git init --bare --shared

which seemed to work for a while, but back to the same permissions error again now. Also tried:

git init --bare --shared="all"

but that didn't work at all.

The solution is to go into the objects dir of the remote repo and manually change the permissions. Not much of a solution. Obviously we are doing something wrong but don't know what it is...it's enough to just start using CVS again.

Short post:
gg=G

Update: It has been pointed out to me that I should set the ODBCINI and ODBCSYSINI to point to somewhere more consistent (eg. /usr/local/etc) from wherever is running the python. So in my shell .zshrc, or in the Apache conf when it's run from the web server.

The other solution I came up with (from actually reading the documentation instead of skimming it), is to use this:

import urllib
import sqlalchemy
from sqlalchemy import create_engine
url = urllib.quote_plus('DSN=STAGEDB;UID=username;PWD=password;DATABASE=MY_DB')
engine = create_engine('mssql+pyodbc:///?odbc_connect=' + url)
engine.connect()

Works a treat.
End update

What a combination...

Had a fair bit of trouble with this one. I'm using the iODBC that ships with Mac OSX and FreeTDS as the driver (that's already up and running for PHP to connect to MSSQL).

Here and here suggest methods to get it installed, none of which worked. Turns out the odbc.ini and odbcinst.ini files have no place in /etc/ or /usr/local/etc on Mac OSX, instead they need to be placed in /Library/ODBC. (Can probably set system wide ODBCINI and ODBCSYSINI env vars but it's a bridge too far at this point...)

The odbcinst.ini is:

[FreeTDS]
Description = FreeTDS Driver
Driver = /usr/local/freetds/lib/libtdsodbc.so

and the odbc.ini (which defines the DSNs) is:

[STAGEDB]
Description = stage
Driver = FreeTDS
Servername = STAGEDB

Ok, so that means this script (just using pyodbc) works fine:

import pyodbc
url = 'DSN=STAGEDB;UID=username;PWD=password;PORT=1433;DATABASE=MY_DB;'
conn = pyodbc.connect(url)

But this doesn't:

import sqlalchemy
from sqlalchemy import create_engine
url = 'mssql://usernamer:password@STAGEDB'
engine = create_engine(url)
engine.connect()

Well, it connects to the database server using the DSN called "STAGEDB" but because the database isn't set, then it's not connecting to a specific database, and you can't simply tell it to connect to a database after this connection has been established (I had a read of the API which didn't help and maybe that's not how db connections are even able to operate – beyond my ken).

So the first solution is to try this in the script instead:

url = 'mssql://username:password@STAGEDB/MY_DB'

But this doesn't work – because we are using pyodbc for the connection, it can only connect by using a DSN (and I can't see how to alter this), and in this case the DSN it's looking for is "STAGEDB/MY_DB" and that's not defined.

So it means I have to define a DSN in the odbc.ini file for every combination of database and database server I want to connect to, for example:

[STAGEDB_MYDB]
Description = stage
Driver = FreeTDS
Servername = STAGEDB
Database = MY_DB

Seems like I'm doing something wrong, but I'm not getting it to work any other way.

This is will a regularly updated post.

First of all, looking at South for db migrations.

Then there is Fabric (like Ruby's Capistrano) for deployment. Not sure we would use this.

For testing ruby (and ruby on rails) has cucumber and rspec, which seem to be pretty close to de-facto standards.

Note: you can't actually have an rspec-like app with python – this is just the way the language doesn't work like ruby. This is a good article about Ruby vs. Python which mentions

For Python/Django the situation seems to be a little more scattered, but I'm intending to use the following equivalents:

• pyccuracy (there is also lettuce) for cucumber
• nose (extension of python's unittest) for (a sort of replacement for) rspec
• The django-nose plugin for using nose when running django tests
• The nose-xcover for using nose with hudson

I'm also going to use hudson for continuous integration (here is a good description about how to get it running with django and nose).

By the way I installed pip for installing python packages and I've started using it instead of easy_install.