[Linda: sorry you got this twice, I accidentally sent it from my home 
address]

Howdy--

So conversations last week in MPK didn't yield anything new or 
startling.  They did, however, reveal that the ON users have
greater frustration and different prioritization than I realized.

If anybody feels a keen desire to own any of this, even just in terms of 
specifying it and/or estimating resources, that's great.  Otherwise, 
I'll give Linda my best estimates.

Prioritization is NOT implied by ordering.

Some things that we currently need:

A. Updates to work with newer versions of Mercurial
    1. Rich did the work and sent me the patch a couple weeks back, I 
need to finish reviewing it and coordinate with Danek and GK.
    2. In addition, Tadd Ottman sent me (unsolicited) a similar patch 
this morning.  I'll compare it to Rich's...

B. Equivalent of "resolve" script to drive Hg merge
    1. We never adopted or worked on David Powell's "hgr" script.
    2. Now Mercurial is actually providing similar support in an 
upcoming release.  (1.1?  I'm not sure.)

C. Automated management of the split repositories
    1. In the spirit of hgext.forest, but need not be generalized.
    2. Major dissatisfier for internal developers, external folks 
probably don't care much, but we might get bonus points if we figure out 
how to sync open source to closed bins tarballs?

D. Text-based merge support
    1. Related to the "resolve" script, but I don't know if Mercurial's 
resolve will include this.

E. Documentation, documentation, documentation
    1. We need something linked from http://onnv.eng/, but it could 
reasonably live outside
    2. Use faq-o-matic or wiki?
    3. Needs to be very basic series of "How do I do XXX?" Q-and-A
    4. Mike (and probably others) have made various updates that are 
either stalled awaiting review, or not prioritized; we should finish 
those, but the one-stop, living document will satisfy most of what I've 
heard or observed as frustrations.
    5. I've been pointing folks to Tom Haynes's blog, because it's 
excellent.  But we really need to suck that information (how to setup 
and run a project gate) into something that we maintain.

F. Outstanding tools bugs
    1. I need to integrate a wad that I've built/started to test.
    2. We need to sweep both bug databases.
    3. We need to capture our ongoing work in RFEs.  (Bugzilla OK)

G. hgsetup
    1. We should consider a cdm-based hgrc check, to advise users of 
newly-recommended settings that we add to hgsetup after they've already 
run it.
    2. Or maybe even making hgsetup smart enough to add to an existing 
hgrc, but that might involve a Python re-write.

--Mark

Reply via email to