Storing information in flat-files has some benefits too.
- Not everyone’s a C[++|#|.Net] programmer - It’s easier to engineer tools in [place you fav. programming/scripting language here] to mainpulate configuration information without having to interface with APIs.
- It’s easier to make changes to another user’s configuration data which is quite an inconvenience with how the various users’ registry sections are stored.
- Some down-sides to using flat-files are
– maintainability - the more flat-files you have, the more you have to remember in terms of filenames, sections within file names, data-types within sections, etc. It’s not to say that registries solve all problems of maintainability, but they sure help.
– scalability - flat-files are quick and easy for a very limited set of data, but they quickly become huge, slow and ugly once you start to store anything-and-everything about your complex, customisable application.
– speed - disk access speeds are relatively slow and top-to-bottom scans of unfixed-length-records in files don’t help when things are already slow.

Registries provide solutions to problems faced with using flat-files but they have their own sets of problems too.
- A collection of configuration information under one roof and providing a consistent API allows for a multitude of tools to be able to search, query, add, modify and delete options with relative ease. Incidentally, backing up and restoring all this information is convenient. A Potential down side is data-loss due to memory or disk corruption on a huge scale (all these eggs exist in one basket).
- Logically gouping options by relavance into hives could allow for easy batch operations - e.g. installing/uninstalling software, services, devices, etc. Backing up and restoring a subset of all your configuration data is also convenient.
- Confgiuring security descriptors on registry objects allows one to control what subsets of access users on the system have in relation to configuration data. Rather than grant a user write-access to a file, you can grant write-access to a set of particular records while still protecting others.
- Generally indexed registries provides better perormance over flat-files in terms of looking up and manipulating data.
- Support for object types allow for effecient storage, organisation and retrieval internally. This to some degree provides filtering of data but is dependant on implementation - e.g. you couldn’t mistakenly store a string in a binary container.

The author forgot the comparison of:

1-(Linux kernel scalabilty and portability)
and (Windows OS/2 kernel)

2-NTFS and (ext3 + swap) file systems
permisions

3-Client/Server architechture
and single application
(what makes Linux really
multitask and multiuser?)

4-The enormous, tremendous flexibility
of the configuration of the UNIX-like OS.

5-The quality of the Open Source applications
is verifiable and is bound to the authors.

I do not know who got the idea to put all the Windows system configuration data and those of the Windows applications into one database file. It would be recommendable to look through patent motions - it’s probable that the same person patented the squared circle or powdered water. The most preposterous idea…

I don’t whether the author was asking a rhetorical question, but the inspiration for the Windows registry was derived from VAX/VMS, since the same person -Jim Alchin- was pretty much the architect and father of both OSs.

That was the same Jim Alchin who sent an e-mail to Gates and Ballmer in which he said: “If I wasn’t a Microsoft employee, I would buy a Macintosh”. Look it up…

-Ramon