Robin discusses his workshop at 23c3. It was succesful (40-100) participants. Suggestions by the public were:
separate the anonimity from the application because it is a very elaborate endeavour. Better use something like TOR as a layer through which the open-search agent works. If you'd make a truly anonymous application it would be very complex to make and there would be a lot of latency. Koen and Erik think however that protocol should be designed with anonymity in mind. We'll elaborate on this in the next meeting.
Make sure search results are consistent. Erik noted that when there will be a lot of data the search results will probably settle in a local minimum, thereby making it somewhat consistent.
There should be an open-search searchbar, preferably as a separate project. An open-search toolbar would facilitate ease of use.
there should be an other name than open-search as A9 (amazon) already got a project called opensearch. Unless we get very big we will never get good rankings in other search-engines + it will be difficult for the general public to disambiguate between the projects. Our suggestion is to give the actual application an other name. (as in Mozilla foundation and firefox application).
Some libraries are not available in C(++)
We should consider also applying for other, easier to remember domains.
Naming conventions
The open-search application will be called an agent. This conveys the multi-agent network it will be part of and gets rid of the client/server distinction as every agent will be both.
we will use a (caching) proxy and a standalone crawler for data collection.
Discussion of the interface draft
Will be elaborated upon and explanations will be written down in architectural design
It is an interactive logo which will change by the progress of the project and the amount of downloads of the agent.
The acticity should be implemented by the end of January, Business cards with the logo should be available before fosdem 07
Other points of discussion
set up a freenode irc channel
central server for e.g. an office through which clients inside that office connect
text vs multimedia: first we will focus on text-based searching but there will be possibilities to have multimedia search plugins.
The architectural design will be written in docbook as that is easily exportable to different formats like pdf, html, etc
The architectural design will be available as a pdf but also as a wikipage per sections with easy commenting options.
There should be different options in the agent to balance secure (recursive search) and fast (iterative search)
We should document our motivations for using/choosing how we will do something. It is also good to shortly describe what we have looked into but didn't use and why we didn't use it. This is also very useful for other people who want to do something similar.
Todo
Robin
set up a freenode irc channel
make a preliminary version of the architectural design documentation
Erik
automate open-search logo, make favicon
look into sourceforge
Koen
look into business cards / flyers
apply for fosdem
Next meeting
brainstorm about anonymity
think about publicity (slashdot, twekers, open-source discussions, set up a source-forge project asking for developers for specific parts of the agent)
discuss preliminary version of the architectural design