~bzr-pqm/bzr/bzr.dev

1185.1.29 by Robert Collins
merge merge tweaks from aaron, which includes latest .dev
1
****************
2
Driving adoption
3
****************
4
5
Getting adoption means persuading people that it's a good choice.
6
7
8
I think the key is to have something that key project leaders see as
9
worth using.  Imagine what it would take to get tridge, havoc, or akpm
10
to switch.  Or not even to switch, but to even just try it out.
11
12
* Simple operations must be simple.
13
14
* The project and the implementation must not have bad smells.
15
16
* Given their current understanding of the problem, there must be at
17
  one feature that's clearly better than what they're currently
18
  using.
19
20
21
What holds it back now?
22
23
* Too complex on initial impression
24
25
* Bad smell from having so many forks/wrappers/kooky opinions
26
27
* Some of the more exotic features can only be appreciated on
28
  familiarity
29
30
* It doesn't actually achieve by default a lot of the advantages that
31
  it ought to: for example it still blocks on the network
32
33
34
Good features at the moment
35
36
* Archive storage is clean; probably makes a favorable impression on
37
  people who look at it
38
39
* Relatively few dependencies (if you don't look too closely at
40
  hackerlab, etc)
41
42
43
From `Ben Collins-Sussman`__
44
45
__ http://www.red-bean.com/sussman/svn-anti-fud.html
46
47
   If you're learning about Subversion and thinking of using it in
48
   your group or company, please approach it the way you'd approach
49
   any new product: with caution. This isn't to say that Subversion is
50
   unreliable... but that doesn't mean you shouldn't use some common
51
   sense either. Don't blindly jump into the deep end without a
52
   test-drive. No user wants a new product forced upon them, and if
53
   you're going to be responsible for administering the system, you
54
   better have some familiarity with it before rolling it out to
55
   everyone. Find a smallish project, and set it up as a "pilot" for
56
   Subversion. Ask for enthusiastic volunteers to test-drive the
57
   experiment. In the end, if Subversion turns out to be a good fit,
58
   you'll have much happier developers (who have been part of the
59
   process from the start) and you'll be ready to support a larger
60
   installation as well. [...]
61
62
   When Subversion hit "alpha" it was already being used by dozens of private developers and shops for real work. Any other project probably would have called the product "1.0" at that point, but we deliberately decided to delay that label as long as possible. Because we're talking managing people's irreplaceable data, the project was extremely conservative about labeling something 1.0. We were aware that many people were waiting for that label before using Subversion, and had very specific expectations about the meaning of that label. So we stuck to that standard. All it takes is one high-profile case of data loss to destroy an SCM's reputation.
63
64
65
`John S. Yates, Jr.`__:
66
67
__ http://lists.gnu.org/archive/html/gnu-arch-users/2004-10/msg00370.html
68
69
  First let me say that I have nothing but increasing respect for
70
  Tom's skills and accomplishments.
71
  
72
  That said I see Gnu Arch as really just emerging from a period
73
  of prototype development.  If the project really wants to take
74
  over the world, and especially supplant projects with momentum
75
  and commercial customers (e.g. Subversion, BitKeeper, etc) then
76
  I would warn against two mistakes I have experienced repeatedly
77
  in my career:
78
  
79
  1) Deferring to a tiny installed base instead of focusing on
80
     eliminating barriers to adoption
81
  
82
  2) Believing that great technology will be irresistible no matter
83
     how it is presented
84
  
85
  Appearances matter.  Expectations matter.  Standards (official
86
  or de facto) matter.
87
88
There is also a `reply from Tom`__.
89
90
__ http://lists.gnu.org/archive/html/gnu-arch-users/2004-10/msg00430.html
91
92
Clear wins
93
----------
94
95
To convince people to use Baz, there has to be some feature they can
96
clearly understand which will be much better under Baz.  It must be
97
something they do today.
98
99
* Offline support
100
101
* Almost no network delays
102
103
* Atomic changes (svn already has this)
104
105
* Correct repeated merges
106
107
* Read-only mirroring archives (not really important) 
108
109
My model is that people will consider changing if 
110
111
1. it's at least as good as cvs/svn
112
2. AND there are no big concerns about implementation/safety
113
3. AND there is at least one feature which is easy to use and a big win
114
115
This gets you to some people at least trying it.  Will people migrate
116
big projects to it?  Maybe, if it looks safe, it fixes there problem,
117
and it doesn't look like something substantially better is on the
118
horizon.