FOSDEM 2011 Oracle. Hmm. Maybe not the favourite word in F/OSS right now ... Unlike Java/OpenJDK/etc – where Oracle has not (yet) dropped the ball – in the LibreOffice camp the cats have left the bags, coops have been vacated, and the code has forked right off ...
Key Libre Office developer Michael Meeks' talk at Fosdem 2011 about prying the (Libre)Office codebase from the grasping hands of corporate control was entertaining, energetic and pointed – and he didn't pull many punches either.
Although it appears that the personal relationships between the developers are OK (LibreOffice and OpenOffice had stalls almost next to one another without any blood being spilled), there appears to be a lot of angst floating around beneath the surface, as with Java – at least to an outsider like me.
While not claiming it as proof, Meeks noted that the number of new contributors leapt by 100-ish when the need to sign a copyright assignment went away, which might simply be a reaction to the removal of tiresome administrative barriers. It could also be a reaction to the inherent asymmetry implied by such agreements, ie: that the licence applies to everyone including the contributor, but not the company, which Meeks says is just not fair or reasonable.
Meeks was scathing about the way such assignments, and hidden constraints such as patents, mean that apparently open-source development can be legally stifled – or at least dangerous.
has invited Oracle
to the party – but
Whether Oracle could stomach accepting Meeks' invitation to join LibreOffice as a peer without special (licence) privileges, or indeed commercially care enough about OpenOffice and LibreOffice at all, is interesting. Oracle staff may follow the fork to LibreOffice if that's where they believe the action is.
Perhaps Oracle would be happy to lose most of the desktop-bound and C/C++ dominated code and replace it with network-safe and server-safe Java(FX), allowing it to take the battle to Google and Microsoft with Oracle Cloud Office? Then Oracle and LibreOffice could coordinate mainly or only on the ODF file formats and a little UI look-and-feel, for example, with face-saving opportunities all round. (A quick scan of the code base suggests less than 10 per cent of the 100kloc is Java.)
Meeks described the efforts now going into cleaning up the code, such as trusting git and so cutting the commented-out CRUD from yesteryear; creating unit tests; and simply getting more people involved in adopting orphaned code.
Meeks added that he wants LibreOffice to have "100 million users", and to support all sorts of slightly obscure uses (such as extracting discs full of teenage poetry from dead word processors as was described in a separate talk), all the while remaining a fun playground for developers.
All this is probably quite a small intersection with areas where Oracle sees the dollar signs, so it could be that – despite the fact that the ideas were clumsily arrived at – the current state of affairs proves to be the best available outcome.
Later in the day, I caught up with Meeks and asked him about the barriers to becoming a "committer" (ie, getting write access to the repository). He said that they are fairly low.
Given the recent Stuxnet and FBI/TCP/BSD code-compromise scares, I asked if he was worried about patient and skilled people being able to plant Trojans in the LibreOffice code-base for example, to which the answer was no, given that there were – in his opinion – bigger live and known-about holes already (maybe the sort of holes that Oracle would not want in the cloud code version?).
Read a FOSDEM interview with Meeks here.
Along with the rabble-rousing in the main venue, there was a large amount of information available on LibreOffice tech – such as how to write format filters (for that angsty teenage poetry) and the new improved build system based on gmake (yes, it seems to have come out of retirement) – and a great deal of internal marketing to developers to make them feel valued.
Certainly if you want to help make the world a better place or just tart up your CV – all while getting your code under 100 million noses – you could do much worse than hack on LibreOffice's code. ®