I've worked with a number of different authors, and there does seem to be a wide range of preferences. I personally like to publish early, and often; I'll often publish a new version integrating just one persons comments (or, of they are nits, batch up a few sets into one version). If there are lots of comments going back and forth, I don't think it is easy for WG participants to mentally keep track of all the different comments, how they interrelate, and what the final text will look like. I also think that it is politer to respond to feedback by integrating and publishing a new version, instead of just saying "Thanks, I'll get to them sometime....". Committing to GitHub kinda accomplishes this, but a: it's harder for participants to find, and b: the github version is a second class citizen. But, other authors seem to have a different view -- they'd much rather get everything fully squared away, all comments addressed, all 't's crossed and 'i's dotted. One of the stickier points is what to do during WGLC -- unfortunately, in many groups this is where the majority of the review and feedback happens, and it is often viewed as poor form to revise during WGLC. It's often hard for the *authors* to keep track of what the consensus is when there are lots of comments, what the new text would look like, etc -- expecting random WG participants to do so is (IMO) unreasonable and leads to frustration and overlapping comments. I'd personally rather publish new versions *during WGLC* saying "This is what this looks like now, does this address your issues?" than trying to explain that we will move text from Section 3.2.5.4.3 bullet 9 to Section 1.7.4 to address Mike's comment, but that will mean that Billy's comments no longer apply because it removes the text that he's commenting on, other than the nits about the case of the acronym, which we agreed to change globally, except in section 8, because it is quoting from another RFC. Confused yet? W On Wed, Mar 1, 2017 at 11:04 AM, Michael Richardson <mcr+ietf@xxxxxxxxxxxx> wrote: > > For the third time in two days I find myself, when asking others for opinions > about some text, pointing at github commit logs. With the beautiful > makefiles we often have, one can't even depend upon having a formatted .txt > version there! > > This is not a rant for or against git or github, but rather about what I > perceive as a shyness about posting intermediate versions of Internet Drafts > to the datatracker. > > I understand that in academia, they never like letting half-baked ideas out, > and so the -00 that we see from academics are often overdue and overly > polished. I know I can't fight that, but at least the -01, etc. could be > issued faster? > > I've even heard some push back from people along the lines of, "wow, that ID > has 27 revisions, is it really stable?", and my feelings have often been more > along the lines of, "wow, that revision has 27 revisions, the authors are > really keen and responsive". > > I appreciate for some reviewers that having more revisions implies that they > think they have to look at the text more often. > > But given the diff utilities, it shouldn't matter how many revisions there > were before times you look, as you can just skip the intermediate versions. > > -- > Michael Richardson <mcr+IETF@xxxxxxxxxxxx>, Sandelman Software Works > -= IPv6 IoT consulting =- > > > -- I don't think the execution is relevant when it was obviously a bad idea in the first place. This is like putting rabid weasels in your pants, and later expressing regret at having chosen those particular rabid weasels and that pair of pants. ---maf