I really hate working with visionaries, most specifically The One Whose Stuff Always Changes. To be more precise, I hate having to base production environments on TOWSAC's ever-morphing APIs and semi-complete implementations of things.
His maxime seems to be "Oh, here's a piece of code which currently works. Let's CHANGE it or its environment or its base or API gratuitously so that it no longer works, but hey, in a few years' time it might do something REALLY VISIONARY".
Besides having some stuff of my own based on his code, I also maintain a load of his code for a variety of purposes, and some of it is really useful. Much of it is badly frayed at the edges, because the fucking NEW VISIONS are always faster to appear than completion of the current stage. The drudgery of finishing implementation and/or documentation of the now never gets a chance over acting on NEW VISIONS. For me, trying to work on top of his APIs, this is a never-ending source of problems and annoyance.
But some of his stuff is quite nice conceptually. So TOWSAC and I tried to cook up a joint paper on something I had contributed a while ago: a mechanism to hide the moving-target-APIs from the end-user by wrapping things into a virtual filesystem. (The push to publish this now came from him, as there's a matching conference in his part of the world with the submission deadline being today.)
I started with coverage of what is, he added what he thinks this might become, things are mostly fine content-wise and so far we have no problems. The problems started when a gratuitous transmogrification of the sensible LaTeX document into Docbook became "necessary". Never mind that Docbook is XML, mostly uneditable and ugly. Never mind that the conference wants submissions in LNCS format (which doesn't cover Docbook), never mind that he knows perfectly well that I dislike XML and that he's perfectly fluent in LaTeX.
The other problem I had with this transmogrification is that it moves me suddenly (a "formatting coincidence" according to TOWSAC, which "could be changed without a problem") over to being secondary author, which I resent (because this stuff is indeed mine from start to finish).
But the final straw is that TOWSAC can't keep from making stupid quips targetting my palace de ork; he's got (IMHO legitimate) grievances and battles with them, but this paper was supposed to be a cooperation between him and me. I am not willing to get pulled into his fight, I don't take any sides but my own.
So we exchanged two somewhat hot emails yesterday and he decided to pulp the mostly-done thing: no joint paper.