[Texmacs-dev] Of teams and coordination

[Miguel de Benito Delgado]

Not so long ago, Joris proposed that we constitute a team of developers 
and interested people. Although not many have shown interest (where are 
you people??!), there is still the possibility that more join in and 
with them there will be an increase in the difficulty of coordinating 
efforts.

During the last week a couple of items in the tracker have been the 
origin of a long series of questions and problems, amounting to an 
absurdly huge waste of time, mainly because it was not clear to me why 
something had to be done (or rather not done) in a particular way. So I 
just kept asking. In the end the problem was that Joris is working in a 
related task ("reorganizations" :-) which will somehow render the 
current approach, and any patchy solutions on top of it like mine, obsolete. 


Most of the pain and waste of time in that exchange could have been 
avoided if either I just shut up and did as I was told (I sometimes 
really should, I'm sorry for that) or if a detailed list of the tasks 
currently undertaken by each developerhad been available. Preparing and 
maintaining this list is in itself a time consuming task whose burden 
falls upon every one of us, but I think the net result would be a gain 
of time.

So I propose the four of us who are actively committing code at the 
moment, and any who are not but anyway work in some part of the project 
or wish to do so, input their objectives into the task tracker in 
savannah. With details, clear scope and current information or it will 
be totally useless. Any ideas or alternative to this approach would be 
very appreciated.

If there is to be such thing as a team, then we must act as one. We must 
know what the others are doing, why and when, or else there is no 
possibility of initiative.

This idea of keeping detailed task lists is evidently most relevant in 
relationship to Joris' work, because he is by far the most active 
developer and the one with the main goals in his head (I realize the 
perceived waste of time my proposal would mean for you, sorry!). The 
approach: "contributors should just prepare patches and send them" is 
fine when the patches address little, focalized problems which do not 
interfere with Joris' current projects, but it's totally inadequate for 
long term subprojects.

Now some ranting...

All this is of course partly in my own interest because I like to peek 
around the code and hate sticking to one particular task. In my opinion, 
cubicle-work is fine for a cubicle-job with corresponding salary and 
increasingly depressing workdays, not for a community. There is 
naturally the opposite view, that we each ought to do what we are told 
to, or what we agreed to in the first place and only under exceptional 
circumstances wander out of our domain. When properly implemented, this 
is probably more efficient in many (most?) cases and what any project 
leader might wish for, but it's not something one usually volunteers 
for. After all, I do this for fun.

Take for example the new "persistent storage". What is its intended use? 
Does that affect work related to preferences? Should I ditch what I did 
with those? And the new buffer related stuff: it renders completely 
obsolete my (few) ideas for better project management (and surely is 
incompatible with some of them). Or take the search feature or... well, 
you get it. The problem is not that my work is ditched because it's 
wrong or can be improved (it always can), the problem is that I 
shouldn't have started it in the first place!

Well, that's enough ranting for an email. Any ideas on the real issue?

Cheers,
Miguel.