On 02/02/2012 04:20 AM, don kretz wrote:
Won't there be an abstraction of some kind for the user interface?
hg is very easy, not harder than zipping and unzipping. There are graphical front-ends. If anybody wants, he can write a front-end talored to PG.
Are all users to be required to get and register an SSH ticket?
All users who want to write to the repository. There will be anonymous read access. Ideally, there will be a hierarchy. The WWers (or whatevers) sit on top and can write into the PG repository. Every WWer can work on his own or have his trusted lieutenants.
Does someone need to download the entire book to fix a comma?
Yes. Once. But since hg transfers are diffed and compressed all subsequent transfers will be very fast. N.B. I don't see hg as a crowdsourcing tool. You can use a crowdsourcing tool (trac ?) on top of hg. But since I think that crowdsourcing is a bad idea, I'll let that as an exercise for people who want to prove that crowdsourcing is good.
How do you get them to avoid the temptation to "really fix this baby up"?
You don't. Same as now. Same as with a web interface. You have the WWers check before they commit.
Who is going to verify the changes? Against what images?
Same as now. You take work from your trusted lieutenants and check it before you commit.
Are the whitewashers pretty familiar with Hg? I at least find it conceptually non-trivial - I'm not sure I'd want to train a non-programmer to use it. I've seen some pretty good developers take a while to get comfortable. But I suppose you've already discussed this pretty thoroughly with everyone who will be affected..
Nope. The first one affected will be me. I'll have to figure many things out as I go. When it works well enough for me I'll offer it to the WWers to use. I'll run this in parallel to the traditional workflow, so nothing changes for anybody except who voluntarily participates to the experiment. -- Marcello Perathoner webmaster@gutenberg.org