1. P2P Bug Tracking with SD
http://syncwith.us
curl fsck.com/sd|perl;
export PATH=~/sd/bin:$PATH;
sd
jesse@bestpractical.com
Jesse Vincent

2. Hi!

3. I’m Jesse (obra)

4. I own a small software
company

5. (Best Practical)

6. I’ve been making issue
trackers since 1995

7. Our software
has some bugs

8. All software
has some bugs

9. (All software
is made of bugs)

10. I spend a lot of time
on airplanes...

11. ...and at conferences
with bad wifi

12. I need to keep track of
our bugs and our work

13. I’m the boss

14. I have no excuse for
not doing my work

15. I need to keep track of
our bugs and our work
even when I don’t have
Internet access

16. I’ve tried everything

17. Text files

18. Text files in
version control

19. IMAP Servers

20. RSS Feeds

21. Running RT on
my laptop

22. Keeping browsers open

23. Nothing was quite right

24. So we built SD

25. SD is a Bug Tracker

26. SD is a Distributed
Bug Tracker

27. SD Features
Tracks bugs P2P Sync
CLI Trac Sync
Web UI RT Sync
Scriptable Hiveminder Sync
Works offline GitHub Sync
Attachments Google Code Sync
Comments Conflict Resolution
Customizable workflow Git integration
Custom properties Darcs integration

28. Principles of distributed
computing

29. The network is reliable

30. Latency is zero

31. Bandwidth is infinite

32. The network is secure

33. Transport cost is zero

34. The network is
homogeneous

35. Topology doesn't
change

36. There is one
administrator

37. Principles of distributed
computing

38. ie s
la c
Principles of distributed
l
F a computing

39. Those are all LIES

40. The network is not
reliable

41. SD isn’t network-
dependent

42. Latency hurts

43. SD runs at the edge

44. Bandwidth is always
a problem

45. SD knows which
changesets you’ve already
seen

46. The network is
insecure

47. SD doesn’t depend on a
network security layer

48. Topology is
unpredictable
and fluid

49. SD is topology-agnostic

50. There is no
administrator

51. (But there are many
people who think
they’re administrators)

52. SD lets you implement
policy in the centralized
systems it syncs to

53. Transport
costs money

54. SD doesn’t use much
bandwidth & can use
non-network substrates

55. The network is
heterogeneous

56. SD is designed to sync to
foreign systems

57. I didn’t make those up

58. Bill Joy, Tom Lyon and
James Gosling did

59. They missed one

60. Data is clean

61. SD doesn’t expect to be
able to enforce data
integrity rules

62. SD runs locally

63. SD plays well
with others

64. It syncs the way you do

65. Clone a project’s bug
database (over HTTP)

66. Work offline

67. Pull changesets from
anyone you work with

68. Publish your database
replica with rsync

69. Topology doesn’t matter...

70. Don’t worry

71. It won’t break

72. SD learns how to
resolve each conflict...

73. ...based on how
everyone else resolves it

74. Using SD (CLI)

75. Getting Started

76. SD Shell
$ sd

77. ./

78. Getting help
$ sd help

80. Creating a new project
$ sd init

82. Project settings
$ sd settings

84. Config file
$ sd config

86. Create a bug
$ sd ticket create

88. Listing bugs
$ sd ticket list

91. Show a bug
$ sd ticket show

93. Update a bug
$ sd ticket update

97. Log
$ sd log

100. Git Integration
$ git sd

102. Using SD (Web)

103. Web? Isn’t SD an
offline tool?

104. Local microserver

105. $ sd browser

106. Home

108. Create a ticket

110. Search

112. Show a bug

114. Update a bug

116. Comments

118. History

120. Working with others

121. Working with others
(Using SD)

122. Any topology

123. It doesn’t matter who
you sync with

124. You get all the updates
eventually

125. Cloning
$ sd clone

126. clone makes a replica of
someone else’s
database

129. Pulling
$ sd pull

130. pull imports unseen
changes from another
database replica

132. Publishing
$ sd publish

133. publish writes out a
copy of your database
replica for sharing

134. (As SD changesets
and static HTML)

136. Hackathon mode
(using Bonjour)

137. Publish your replica
$ sd server

138. Pull updates
$ sd pull --local

139. Working with others
(Using other systems)

140. But you already have a
bug tracker?

141. No Problem!

142. I use at least
two others.

143. I wrote at least
two others.

144. We designed SD talk to
foreign bug trackers

145. RT

146. (RT::Client::REST)

147. Hiveminder

148. (Net::Jifty)

149. Trac

150. (Net::Trac)

151. Google Code

152. (Net::Google::Code)

153. GitHub

154. (Net::GitHub)

155. Redmine
(Read-only for now)

156. (Net::Redmine)

157. Want to help with
Bugzilla, Jira, FogBugz or
something else?

158. Work with
foreign
replicas in
a Star
Topology

159. X
An SD node
can act as a
gateway

160. Clone
$ sd clone

161. Clone from Google Code
$ sd clone --from gcode:k9mail

162. Clone from RT
$ sd clone --from "rt:https://rt.cpan.org|DBI|"

163. Clone from Trac
$ sd clone --from trac:https://trac.parrot.org/parrot

164. Clone from GitHub
$ git sd clone --from github:miyagawa/remedie

165. SD reverse engineers a
database history

166. Pull
$ sd pull

167. SD reverse engineers a
partial database history

168. Push
$ sd push

169. SD figures out local
updates and sends
them upstream

170. (Then it does a bunch
of book-keeping)

171. So what’s that all look
like in the real world?

172. Google Code Alice’s SD Bob’s SD
Create some bugs
clone --from gcode:k9mail Alice’s SD makes a full copy of the k9mail from google code
Someone updates ticket 12
ticket update 13
ticket create
pull --from k9mail SD integrates upstream changes from Google Code since
Alice ran the ‘clone’ command.
push --to k9mail SD figures out what’s changed locally since the original
‘clone’ command. SD then sends any changes that Google
Code hasn’t seen upstream as ticket updates. Alice’s new
ticket and her updates to ticket 13 are now part of the
upstream bug database in Google Code
publish --to alice.com:public_html/k9 SD exports a full copy of Alice’s bug database and rsyncs it
up to alice.com/k9. Anyone browsing alice.com/k9 can clone
her SD replica. (Or see it in static HTML)
clone --from http://alice.com/k9 Bob makes a full copy of Alice’s SD replica of the k9mail
database. He’s now free to make changes to his local
database replica.
ticket update 45
ticket create
ticket create
ticket update 45
publish --to bob.com:public_html/k9 Bob publishes his full K9 bug database. Anyone browsing
bob.com/k9 can clone Bob’s K9 replica. Anyone who already
has a copy of Alice or Bob’s database can incorporate Bob’s
new ticket and his changes to ticket 45.
pull --from http://bob.com/k9 Alice’s SD pulls all previously unseen updates from Bob’s
published database replica and incorporates them. If Bob's
edits to ticket 45 didn't conflict, Alice's SD incorporate them.
If they did conflict, Alice will be prompted to resolve the
conflict.
push --to k9mail Alice publishes all updates her bug database has seen since
she last sent updates to Google Code. Google Code now has
all of Alice and Bob’s changes.

173. Installing SD

174. It’s time to get SD
up and running

175. SD is in Perl

176. SD uses 45-90
CPAN modules

178. Are you afraid?

179. Are you afraid?

180. Don’t be

181. I have a novel idea for
a Perl application

182. One-tweet install.

183. I’m installing #SD (http://syncwith.us)
after seeing @obra’s talk at #OSCON!

184. 80 bytes! I still have 60
to work with!

185. I’m installing #SD (http://syncwith.us)
after seeing @obra’s talk at #OSCON!
curl fsck.com/sd|perl;
export PATH=~/sd/bin:$PATH; sd

186. You’ll need:
curl, perl 5.8, C compiler

187. You won’t need:
CPAN

188. You won’t need:
to answer prompts

189. You won’t need:
to fix dependencies

190. SD is in Perl

191. SD uses CPAN
modules.

192. This is a blessing.

193. This is a curse.

194. CPAN
=
Dependency Hell

195. When we first built
SD, we used anything
we thought was useful.

196. (The first version of SD
used a SVN backend)

197. The first ~useful
version of SD:

199. 123 Dependencies

200. ...a couple hours later

202. 54 Dependencies

203. Only one needs a
compiler.

204. Shipwright gives us
one-command install

205. curl fsck.com/sd|perl
export PATH=$PATH:~/sd/bin

206. What’s that do?

207. curl

208. perl

209. $ head /Library/WebServer/Documents/sd
open (my $tar,'|tar xz 2>/dev/null');
while (<DATA>) {
print $tar $_;
}
close $tar;
exec("cd sd-build; bin/shipwright-builder
--install-base=$ENV{HOME}/sd");
__DATA__
?I?Isd-build.tar?<is?F???_1?䀌yH?"#‫{$ۿ‬U???H?eH
...

210. Shipwright installs a few
Perl modules...

212. ...in order

213. Scalar-List-Utils WWW-Mechanize-GZip Sub-Exporter
String-BufferStack File-Slurp Path-Dispatcher
Class-Accessor Test-MockModule Module-Pluggable
Class-Data-Inheritable Net-GitHub Time-Progress
Tree-DAG_Node MIME-Types Carp-Assert
Test-Simple Class-Singleton Proc-InvokeEditor
Sub-Uplevel Params-Validate Test-HTTP-Server-Simple
Test-Exception version Module-Refresh
Array-Compare ExtUtils-CBuilder Carp-Assert-More
Test-Warn ExtUtils-ParseXS Test-LongString
Template-Declare Test-Harness Test-WWW-Mechanize
URI File-Temp Test-Script-Run
HTTP-Server-Simple Module-Build prophet.git
Params-Util DateTime-TimeZone Devel-StackTrace
Class-Inspector List-MoreUtils Exception-Class
File-ShareDir DateTime-Locale Error
DBI DateTime RT-Client-REST
DBD-SQLite File-MMagic Email-Address
HTML-Tagset Net-Google-Code YAML
HTML-Parser Term-ReadLine-Perl Path-Class
HTML-Tree Digest-SHA1 Clone
Crypt-SSLeay Digest-HMAC Hash-Merge
JSON Net-IP UNIVERSAL-isa
YAML-Syck Net-DNS UNIVERSAL-can
JSON-XS Net-Bonjour Test-MockObject
JSON-DWIW TermReadKey Net-Jifty
JSON-Any Data-UUID Lingua-EN-Inflect
Mouse XML-Atom-SimpleFeed Text-CSV
Any-Moose Digest-SHA Net-Trac
Compress-Raw-Zlib Exporter-Lite boolean
Compress-Raw-Bzip2 IPC-Run3 Time-Piece
IO-Compress MIME-Base64-URLSafe Test-MockTime
libwww-perl Data-UUID-Base64URLSafe DateTime-Format-Natural
HTTP-Response-Encoding Sub-Install sd.git
WWW-Mechanize Data-OptList

214. (104 dists with our
sync plugins)

215. What’s wrong with SD?

216. $ sd clone --from http://fsck.com/~jesse/sd-bugs

217. Don’t want to
install SD?

218. Point your browser at
http://fsck.com/~jesse/sd-bugs/html

219. What’s next?

220. Indexing

221. GPG-signed changesets

222. Actually releasing 1.0

223. Thanks!
Questions?
curl fsck.com/sd | perl
http://syncwith.us
jesse@bestpractical.com