From: Cristian B. <cr...@cs...> - 2008-11-02 15:26:30
|
Manuel GAY wrote: > because some people in BEST were having makumba parties, some makumba > developers went there, and thus the name. > > not sure this is suitable for a thesis though :) > > cristi, any more insights, regarding the first question also? not much more, in 1999 when we started we had another name (metadata, known in athens summercourse 1999 as metafeta) but by the GA 2000 (i think) we thought that name is not best-specific-enough. i think it was a discussion between Igor Borojevic, Stefan Baebler and myself, not sure who else took part. Though I might confuse that discussion with the discussion where Igor (I think) baptised Karamba. or it may be that they were "branded" at about the same time, which in retrospect looks like a smart move. at the time we had Minerva which was more or less done in makumba (but only the DB layer survived from that makumba version, the rest was some non-jsp templating, i think JHTML was involved, see below), PA which was in Lotus Notes and Johnny which was a java application with a "avant la lettre" JSP-like language that I and Ovidiu developed since 1996 (it was called JML, Sun's JHTML appeared after, based on a product called ATG Dynamo, and JSP appeared around 2000 as a marketing response to ASP) and persisted its state as Java serialization. I imported data from all those in Makumba in a painful 2001 summer (i had planned 2 days for it, it took 2 months). so can you imagine that BEST committed to a makumba that was more or less just an idea, had no mak:list and not to mention mak:form in 2000? karamba design discussions (workshop and GM workgroups) were well under way already in 2000. Stefan stayed in Stockholm 6 months autumn/spring 2000/2001 working on implementing both the makumba and the karamba concepts, as well as joining these karamba working groups. we also made the bundle then. i think after Stefan left came the idea of implementing makumba as a JSP taglib, so when i did the data import in the summer 2001 i already had mak:list to look at it. i designed and implemented mak:form and BL in autumn 2001 in Ivrea at a summerschool. finally we had a complete makumba. i don't know when we switched to tomcat, i guess ever since we worked with jsp. before that, there was something called Java Web Server and it was the reference implementation for JHTML. then ITC started the implementation of PA TS, Minerva TS and Johnny TS (= transitional system, i.e. a makumba clone of the original). parade was developed quite fast in early 2002 and launched during a ITC meeting (Risto called it paralyzed development). i think Johnny TS (which was crucial for the success of the whole mak+karamba enterprise) was put in production in summer 2002, Raul started it and went home, then it turned out that there had been an outofmemory exception when running the import that i had developed already a year before, so lots of people and their applications were missing. we did a re-import and a handful of applications (entered before we realized that the import was wrong) were lost, but overall it was ok. just in time because johnny had not been designed to be multi-season and it was caching all its data, so we talk 1 Gb RAM (there were no Java weak references at the time) and 1-2 hours to start up. i don't remember when PA-TS and Minerva TS were launched. then there was one project left, the account merging, that took much longer to complete. if you want to find out the precise "makumba" moment and circumstances, maybe ask the above people, and Toto. see also the list of initial makumba meetings http://www.makumba.org/makumba.html ... so the makumba.org domain name was bought in summer/autumn 2000 but the package name remained metadata until late 2000. if i look at that document now it seems that we were in the confused design space exploration that lasted quite long (over 2 years!) before the mak:list and mak:form breakthroughs were made. cristi |