> On 13 Feb 2015, at 4:14 pm, Dennis E. Hamilton <[email protected]> > wrote: > > There is a great lessons-learned article at > <http://jasonpunyon.com/blog/2015/02/12/providence-failure-is-always-an-option/>. > > Joel Spolski, whose company supports Stack Exchange, made this comment in his > twitter message about it: > > "Now you know why experienced developers are really conservative about > technology choices." > > My reason for remarking on this, beside it being a great story, is the "I'd > Really Like This Done By Friday" topic on the very end. I notice that we do > not seem to have very many places in Corinthia, right now, where such > prospects for agile baby-steps stand out. (The request to make Windows > Native scripts for the externals downloads and extractions is the closest > I've seen that I could act on that way.)
Anyone can start writing a filter today, without immediate concern as to ensuring it fits within an a particular architectural structure. It doesn’t even need to use any of the functions in the library - if the algorithms are in place, they can be subsequently integrated into the main codebase. Here’s a couple of other quotes from the article: "Bad planning, bad tech decisions, and a propensity for sinking weeks on things only incidental to the problem all adds up to excruciatingly…slow…failure.” "We dropped back to the simplest system we could think of that would get us going" ... "Within two weeks we had real-world tests going and we got incredibly lucky. Results of the model tests were nearly universally positive.” My takeaway from this is that it’s a bad idea to obsess too much over architecture, build processes, and getting a perfect underlying structure before starting on feature work that matters to end users. I think this an extremely valuable lesson. > I am thinking that moving Corinthia to Apache might have been a bit premature > with regard to the state of the code base and the degree to which the > super-programmer has ideas about how it is to become different. Agreed, > Corinthia is a podling. However, it seems me that the reason for being an > Incubator project is not so much about the status of the code, it is more > about a place to transform to a project and process that evolves toward some > level of *sustainable* maturity. Currently, there is a lot of moving target > and it depends on the super-programmer as benevolent dictator at this point. Well, I don’t want to be a benevolent dictator. Sure, I have a lot of ideas for the project - as do others - that I’d like to see come to fruition some day. But my ideas shouldn’t carry any more weight than anyone else’s. Jan is another person who has lots of ideas and has done significant work on improving the architecture (prior to his involvement there was no clean separation of concerns in the codebase), and I anticipate further improvements in this area. But I disagree that this means it was too early to move into incubator, or that it’s too early for people to start working on whatever they are interested in the project. As I mentioned previously, if we wait for a stable base then nothing will ever get done, because the base is something will evolve to meet our needs as a team, and that’s only something that can be done with involvement from all of us, working together. My “out-there" ideas about a generic parser library, the statically typed, functional programming language for implementing language transformations, the implementations of filters in that language as opposed to C, formal verification of transformation algorithms etc are all just stuff I’ve been thinking about. The last thing I want is for these or any other suggestions I’ve made to be seen as reasons for delaying beginning of coding work on stuff like ODF. > With regard to exhortations to start doing something, I just stumbled on the > quotation at the beginning of Mythical Man Month Chapter 6, Passing the Word. > > He'll sit here and he'll say, "Do this! Do that!" > And nothing will happen. There was a bug report someone filed about requesting a roadmap. I was waiting for someone else to fill one in, but after a few weeks of nothing happening on that front I came along with my comments about ODF (which I assume you’re referring to). My suggestion really comes down to the notion that we should be focusing on new features or major improvements. I’m well aware of the difficulties caused by lack of proper documentation - which I’m continually trying to find time to get around to. I have to juggle this with paying work though which is keeping me quite busy right now. However, I’d love to answer questions about “How does thing X work?”, “Why did you do Y this way?” etc. as I think this would help people along and also be a good indication of the most important priorities for documentation. — Dr Peter M. Kelly [email protected] PGP key: http://www.kellypmk.net/pgp-key <http://www.kellypmk.net/pgp-key> (fingerprint 5435 6718 59F0 DD1F BFA0 5E46 2523 BAA1 44AE 2966)
