Forum OpenACS Q&A: Contacts package

Since contacts package seems fuck up installations big time I decided to create a seperate topic on it. Here's my impression for starters. Developers, feel free to add comments and keep us up to date of developments. I know you post them here, but that information doesn't realy indicate if functions will realy work.

Current state of affairs:

1. Contacts is not tagged as a mature package and it will messs up all accounting related modules that depend on it. Unfortunately the dependancies are not clearly marked, so you may run into crashes if you decide to give it a try.

2. Can we mark contacts a priority, and set a deadline by which it needs to be fully functional.

3. Who's working on getting this sorted out. I see Malte and Thorben being involved in the topic. Are there other takers? Is there anything you folks need assistance with? I am sure some of the new kids on the block could help with moving forward.

4. Contacts relates to addresbook. Since that package has been depreciated 4 years ago, please remove it from the tarball. It's broken, and nobody uses it.

Setting a deadline and marking a software with a priority is not how an open source community works. I am aware that a lot of interest exists in contacts and that people would like to use it for their various purposes, but as a developer who extended contacts I can tell you that you have to be well trained and knowledgable of the pitfalls of the system to be able to use it. You can by choosing an ill advised combination of Expert Searches (which you can run on your own on your site) crash your system within seconds. To setup PDF generation for mass mailings you need to be really knowledgable about the usage of Linux, X-Windows, VNCServer, OpenOffice and the OpenOffice XML standard. This is just to get it running.

Furthermore you need to be trained and fully understand how groups, relationships and attributes work together.

I guess my main point I'm trying to make it: Do not expect a package that you can install and it "just works". It is more complex than assessment and with the callbacks and specific integrations into other packages it is only getting more complex.

As for 3). Anyone is free to help out. This is why we work out of the openacs.org repository.

Thanks for the explanation Malte. It helps newbies to know these things. We'll probably need a topic on the wiki about this aspect.

Thank's for all the energy you folks put into creating the tools.

This all may be true Malte but other developers cannot just jump in if the software is never stable enough to install cleanly. You need to at least produce a developer release if you expect others to get on board. It's unreleastic to expect someone to download a mess out of CVS that will take days to sort out.

Michael, fundamental question. Would you prefer us to not work out of openacs.org CVS unless we produce a developer release?

There might be some misunderstanding, but at the current point I do not expect anyone to use contacts out of HEAD or get on board for that matter. Use the version in 5.1, works fine to my knowledge.

Others reported they were able to install the version from HEAD, so I think the statement of "mess" "that will take days to sort out" is just unjustified.

In the future we luckily wont have these problems anymore as only tagged packages will show up in the repository and therefore only users who download packages from CVS will see out code (until we release it). This should prevent this whole discussion from happening in the first place.

Malte, to be honest now that xowiki is starting to rev up (check out the categorized/collapsable view dave setup!), the next issue that needs to be dealt with is the need to define the testing process.

I see the testing process as two separate stages:

1. clean install / uninstall
2. feature testing

However within those two stages there are a lot of variable possibilities, and I think that requires some input from all the devs.

The testing process should be clear and standardized (if possible) so that we aren't basing results on arbitrary criteria.

I haveto first fix some mis categorized documents in the wiki, then I will put up a testing process document where you guys can hopefully come in and streamline the process.