[OSCOM] content import/export sprint?

bergie and i had a little chat about content import / export, and we 
thought this might be of general interest. would there be an interest in 
a sprint around this?

[11:36] bergie: gregor: BTW, we've made first production-level Exorcism 
this week
[11:36] gregor: cool!
[11:36] gregor: and i see you are having fun
[11:36] gregor: rich bowen seemed to be very impressed ;)
[11:36] bergie: 1) Read database of a proprietary CMS using JDBC, and 
mangle the content into XML format
[11:37] bergie: 2) Modify the XML to Midgard's needs
[11:37] bergie: 3) Import via JCR
[11:37] bergie: Jukka estimated CMS export plugins should be about 500 
EUR per CMS, so they can't be much work. Haven't looked at the code yet
[11:39] bergie: ...of course, that work estimate is based on finnish 
typical CMSs, which are not really complex
[11:42] gregor: all those boutique cms,eh
[11:44] bergie: yep. I think this has quite a huge potential
[11:45] gregor: indeed. consolidation is good
[11:45] bergie: in this exorcist run we got all content, navigational 
structure and users out of 
http://www.emedia.fi/cgi-bin/iisi3.pl?cid=www&mid=148&sid=170
[11:46] bergie: so only things you need to convert manually are layout, 
permissions, and possible custom forms etc
[11:46] bergie: To ensure fairness we probably need to develop Midgard 
export filter too :-)
[11:47] bergie: then if we have filters for some OS CMSs, we could 
switch the OSCOM site between systems every week ;-)
[11:47] gregor: lol
[11:47] bergie: does Lenya support JCR yet?
[11:48] gregor: partially
[11:48] gregor: the site structure can be stored in jcr, content not yet
[11:48] bergie: I think once we get Exorcist into a "Project" stage with 
some docs, etc, we could consider running an Exorcist Sprint
[11:49] gregor: that would be very valuable
[11:49] bergie: though I guess not many OS CMSs support JCR yet... 
Midgard, Magnolia and Lenya?
[11:49] bergie: and in Midgard it isn't available yet in production release
[11:50] gregor: i guess we'd write more import filters (they are 
cms-independent, no?)
[11:51] bergie: well, you need to map the data to the particular CMSs 
data structure
[11:52] bergie: so it will be interesting to see how this scales cross-CMS
[11:52] gregor: <some cms> -> <JCR> -> <yourcms> no?
[11:52] gregor: the first -> could be shared, no?
[11:53] bergie: 
http://snip.yukatan.fi/space/start/2005-02-21/1#CMS_migration_with_the_Exorcist
[11:57] gregor: yeah, what i am wondering is.. if we both wanted to have 
importers for typo3,say, could we share the export plugin from typo3/
[11:57] bergie: yep
[11:57] bergie: I think so
[11:58] gregor: hopefully
[11:58] bergie: need to check the current status from Jukka though
[11:58] gregor: then a sprint would make sense
[11:58] bergie: but shared export filters would really make sense here
[11:58] gregor: otherwise you have a m*n explosion
[11:58] bergie: yep, would be cool to have import/export with all 
participating CMSs
[11:58] gregor: combinatorial number of plugins required, is what i meant
[11:59] bergie: yeah
[11:59] bergie: well, the way Jukka described it in his blog, it used 
generic export format, then XSL transformation into specific format
[12:00] gregor: right
[12:01] bergie: so each CMS needs only import and export plugins, but 
the transformation is specific to the combo
[12:01] bergie: i.e. you'll need Typo3->Lenya transformation and 
Typo3->Midgard transformation
[12:02] bergie: because somehow you need to map the concepts of 
different CMSs together, and there is no generic enough format for that
[12:02] gregor: yeah, too bad
[12:02] gregor: so it remains a point to point effort
[12:02] bergie: the transformations do, but getting the export/import 
plugins into CMSs would the big first step
[12:03] gregor: that is true
[12:03] bergie: of course, if you don't want to be point-to-point, you 
could use a format like PSI http://psilib.sourceforge.net/, but then you 
probably lose data you could get through specific transformers
[12:03] gregor: and the transform is always lossy in some ways
[12:03] bergie: yeah, depends a lot on what the CMSs support
[12:03] gregor: so the question is, how much is acceptable
[12:06] bergie: there was slight data loss in the eMedia->Midgard 
conversion too... we lost permissions because they used a really cryptic 
storage format for them that we didn't have time to figure out
[12:06] bergie: I think in most cases clients will be really happy 
already with getting their site structure and articles through
[12:07] bergie: even better if you can map pages to "types" like news 
and events correctly so that the imported site will basically work
[12:08] gregor: yep