On Tue, 2015-11-03 at 16:14 +0100, Ximin Luo wrote: > On 03/11/15 15:57, Ian wrote: > > It's interesting that you would point to the Hurd as an example of > > how a > > software project should be run, perhaps if they had spent more time > > on code > > and less on documentation, as Linux did, Linux wouldn't have > > utterly won > > that battle. The GNU Hurd is a cautionary tale, not a model for > > software > > development. > > > > As for your belief that documentation will solve everything (as it > > clearly > > didn't for GNU Hurd), you may not be aware of the "Agile > > manifesto", which > > has become the dominant project management methodology over the > > past 15 > > years or so. On of the tenants of this is "Working software over > > comprehensive documentation". > > > > You're applying a generalisation inappropriately here. Linux did not > win because they spent time writing code instead of documentation. > They did both. And even now they're taking ideas from Hurd and > microkernel theory. > > The Agile manifesto is a load of crap. I've seen more teams and more > projects destroy themselves by deferring to the Agile manifesto to > excuse bad engineering, than destroy themselves by overengineering > and overdocumentation. > > Also Bob was talking about the current Hurd project. It's a stretch > to say that Hurd lost "because" of overdocumentation. There were > technical problems with microkernel architecture that were not solved > for a long time, less people in FOSS understood it, blah blah blah. > Some microkernels like L4 are becoming more succesful today, and in > the long run they have inherent advantages over monolithic kernels, > such as being easier to formally verify. > > > You may also be unaware that in the ideal case, well-written code > > is > > self-documenting, making external documentation unnecessary. This > > is one > > of the tenants of the "clean code" methodology. > > This is for *other developers*, not for external reviewers or other > projects that might to interoperate with different components of your > product. It's irrelevant to what's being discussed. > > FWIW, I don't know if Hurd is a good example to follow for Freenet. I > haven't reviewed their documentation yet. It's just this thread has > very bad points in it from both sides. And the counterarguments are > all missing the point. Take a step back to see the wider picture. >
Tor and its ecosystem is a better example... Florent
signature.asc
Description: This is a digitally signed message part
_______________________________________________ Devl mailing list Devl@freenetproject.org https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
