It's an old fallacy. The costs of maintaining, modifying, enhancing, and testing your own code as well as the time lost that could be used on other aspects of the system (opportunity cost) typically exceed the costs of doing all that you've said.

A good, automated system to generate installs sets & scripts and actually do the installation goes along way.

If you have thousands of pages of documentation, something is wrong.


>It is one additional file we have to document, maintain from servers to servers, add references in the project, verify for licensing issues, replicate in the migration procedures and etc, etc., etc. On a nationwide project involving thousands of documentation pages, dozens of servers and a huge infrastructure, I think it is easy to understand that bringing things to the lowest common denominator would make sense.