Edmund Wong wrote:
> Hi,
>
> I've been giving this 'patch' some thought and am wondering if it
> was necessary.
>
> I think the original intention of the functional specification post
> was to become a RFC. Would saving the functional spec to a
> text file and placing it in 'notes/' directory defeat its purpose as a
> RFC text since users might want to give their $0.02 without
> needing to checkout some text file?
It's no good asking lots of people to suggest quick comments on an empty
"spec": as you have discovered, lots of people have contributed lots of
comments to the discussions already. The difficulty is organizing those
thoughts and comments into a nice, tidy, consistent, understandable,
maintainable, learnable, user-friendly, configurable, and
chocolate-flavoured form.
What it needs is for one person to think hard about how to make a
feature that does most of what most people want it to do, without
leaving messy edge cases, inconsistent or unintuitive behaviour, and so
on.
Once you (or whoever) have written a spec that is your best effort
attempts to define such a behaviour, then we can ask everyone to
criticize it (look for inconsistencies, etc.) and then you can try to
fix the problems. When people generally agree that it's a good spec then
we say "OK, that's agreed by consensus, so we can do it."
When you have a first draft that's ready for soliciting comments (an
RFC), that would be a good time to commit it. (I would suggest a
subdirectory as there will likely be more than one doc.)
- Julian
> What's (for lack of a better word) 'bothering' me about this is
> the fact that there are a lot of comments as given in the link
> (http://svn.haxx.se/users/archive-2008-03/0462.shtml)
> and then there are the archives which needs sifting through.
> So a RFC isn't necessary.
>
> As I went through part of the linked text, I noticed that a lot
> of the discussions (including the patches as submitted by P.Marek)
> were inconclusive. I don't know, but I feel that the submitted patches
> might prove to be 'unpatchable' given the fact that it was done against
> the 1.1.1-era trunk.
>
> Perhaps a RFC is necessary right now, if not for comments but
> to conclude what has been discussed in the past and give those
> discussions some closure so that implementation of some sort can
> start(RFC now being "request for conclusions"). Then again, perhaps
> there are new 'ideas' to sort through.
------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2417513
Received on 2009-11-13 12:31:40 CET