I'd like to make a few comments on the thesis document for the Alerts system (I'm not commenting on the code/package itself because I haven't looked at it, but it is a welcome addition).
The author of the thesis document (Andrew Zeon) says the following:
"One of the disadvantages of OpenACS is the complexity of installation. Installation of the AOLserver, PostgreSQL database, and the OpenACS toolkit are not done using graphical wizards but rather it requires the installer to manually type in every command which means that installation may only be carried out by a Linux/Unix professional"
It is true that OpenACS requires several pieces to work, and these have to work together. However, _any_ web framework of the same port of OpenACS requires just as many or more pieces. Try installing and setting up Jakarta+JBoss+Apache+Mod_jk+PostgreSQL.
OpenACS installation does _not_ requires the installer to "manually type in every command". That is _one_ way you can go about doing it.
PostgreSQL has packages available for it for all major Linux distributions, other UNIX systems, and Windows. AOLserver does as well.
The only piece that you really have to install "by hand" is OpenACS, and even that is being worked on. Not everybody is fond of "graphical wizards" for installation of software, nor does it mean that they make life any easier (download PocketMoney for the Palm OS for an example. It has a "graphical wizard", and that doesn't make installation any easier).
Further, the author says:
"Another disadvantage is due to the nature of open source projects, the code inside the packages are usually messy and hard to understand. A great deal of time must be spent on trying to understand the code as there are are only limit amounts of documentation and support"
The messy and hard-to-read parts of OpenACS (there are several of those) has _nothing_ to do with its open sourceness. Absolutely _nothing_.
I would doubt that Andrew has ever worked in a proprietary software. He would see that the messiness and hard-to-read parts are there as well. One just has to look for Microsoft to see that in full-fledged practice.
What the open sourceness of OpenACS gives the project is the opportunity for people like Andrew, a student trying to learn (and anyone else), to see, learn from and correct such problems. You would _never_ have this opportunity with a proprietary solution, with very very few exceptional cases.
I would also argue that there's plenty of support for OpenACS. There are a dozen companies giving support for a fee and hundreds of users giving support for free right here on the forums. Support that Andrew has made use of right here on these forums.
So the hard-to-read and messy code, along with the lack of better documentation are indeed problems. Not the open sourceness of the product.
-Roberto
