5 - --retry option for install, so it won't complain about a directory already
7 - The calling web code invocations are a mess, with stubs living
8 in the install, deploy modules and the real deal living in util. Furthermore,
9 we use the scripts-specific heuristic to determine where the app
10 lives, and the only reason my test scripts work is because they
11 get manually fed the domain and path by my environment variables.
13 We will record the URL used for the initial installation, and save it in
14 .scripts/url. If autodetection in either direction is
15 available, we verify this value against the actual file path the installation
16 lives in (for the scripts case, we can do a file-level comparison because we
17 know the web root of any given file). If they mismatch, we error out
18 and have someone manually resolve the problem. If autodetection is not
19 available, we use the saved .scripts/url for operations.
21 - Test code duplication (app specific parameters), also naming sucks for autocomplete
22 - wizard install wordpress should ask for password
23 - Test code should auto-nuke the database using `wizard remove` before doing a new install
24 - git diff :1:$file :2:$file to find out what the user did, or is it :3:?
25 - Document how to fix a broken upgrade
26 - php.ini needs to get substituted!
27 - Make wizard install accept appname-head (so that you can do a test with
28 head, and do things without tags). Also make it accept commit hashes.
29 In fact, let it accept any committish. Figure out what to do if we
30 do a test script with x.y.z when we REALLY mean x.y.z-scripts. XXX!!!
31 - Do early validation of inputs for configuration
32 - Let 'wizard configure' be interactive
33 - Parse output HTML for class="error" and give those errors back to the user,
34 then boot them back into configure
35 - Get rid of our custom sizing code and use dialog's built-in sizing (i.e. width=0, height=0).
36 Maybe our sizing code is superior, maybe not.
38 - Replace gaierror with a more descriptive name (this is a DNS error)
40 - Pre-emptively check if daemon/scripts-security-upd
41 is not on scripts-security-upd list (/mit/moira/bin/blanche)
43 - Redo Wordpress conversion, with an eye for automating everything
44 possible (such as downloading the tarball and unpacking)
46 - Web application for installing autoinstalls has a hard problem
47 with credentials (as well as installations that are not conducted
48 on an Athena machine.) Possible solutions include asking the user
49 to SSH into an athena machine and run a bunch of commands, or writing
50 a Java applet (possibly in Clojure or Scala) which gets filesystem
51 permissions and then performs the operations.
54 - Genericize callAsUser and drop_priviledges in shell
55 - Summary script should be more machine friendly, and should not
56 output summary charts when I increase specificity
57 - Summary script should do something intelligent when distinguishing
58 between old-style and new-style installs
59 - Report code in wizard/command/__init__.py is ugly as sin. Also,
60 the Report object should operate at a higher level of abstraction
61 so we don't have to manually increment fails. (in fact, that should
62 probably be called something different). The by-percent errors should
64 - Move resolutions in mediawiki.py to a text file? (the parsing overhead
66 - If a process is C-ced, it can result in a upgrade that has
67 an updated filesystem but not updated database. Make this more
69 - PHP end of file allows omitted semicolon, can result in parse error
70 if merge resolutions aren't careful. `php -l` can be a quick stopgap
73 - Figure out why Sphinx sometimes fails to crossref :func: but wil
74 crossref :meth:, even though the dest is very clearly a function.
75 Example: :func:`wizard.app.php.re_var`
76 - The TODO extension for Sphinx doesn't properly force a full-rebuild
78 - Make single user mass-migrate work when not logged in as root
79 - Don't use the scripts heuristics unless we're on scripts with the
80 AFS patch. Check with `fs sysname`
81 - Make 'wizard summary' generate nice pretty graphs of installs by date
82 (more histograms, will need to check actual .scripts-version files.)
83 - It should be able to handle installs like Django where there's a component
84 that gets installed in web_scripts and another directory that gets installed
86 - ACLs is a starting point for sending mail to users, but it has
87 several failure modes:
88 - Old maintainers who don't care who are still on the ACL
89 - Private AFS groups that aren't mailing lists and that we
91 A question is whether or not sending mail actually helps us:
92 many users will probably have to come back to us for help; many
93 other users won't care.
95 PULLING OUT CONFIGURATION FILES IN AN AUTOMATED MANNER
97 advancedpoll: Template file to fill out
98 django: Noodles of template files
99 gallery2: Multistage install process
100 joomla: Template file
101 mediawiki: One-step install process
102 phpbb: Multistage install process
103 phpical: Template file
106 wordpress: Multistage install process
108 COMMIT MESSAGE FIELDS:
110 Installed-by: username@hostname
111 Pre-commit-by: Real Name <username@mit.edu>
112 Upgraded-by: Real Name <username@mit.edu>
113 Migrated-by: Real Name <username@mit.edu>
114 Wizard-revision: abcdef1234567890
115 Wizard-args: /wizard/bin/wizard foo bar baz
119 Committer: Real Name <username@mit.edu>
120 Author: lockername locker <lockername@scripts.mit.edu>
124 - It is not required nor expected for update scripts to exist for all
125 intervening versions that were present pre-migration; only for it
126 to work on the most recent migration.
128 - Currently all repositories are initialized with --shared, which
129 means they have basically ~no space footprint. However, it
130 also means that /mit/scripts/wizard/srv MUST NOT lose revs after
135 * Some parts of the infrastructure will not be touched, although I plan
136 on documenting them. Specifically, we will be keeping:
138 - parallel-find.pl, and the resulting
139 /mit/scripts/.htaccess/scripts/sec-tools/store/scriptslist
141 * The new procedure for generating an update is as follows:
142 (check out the mass-migration instructions for something in this spirit,
143 although uglier in some ways; A indicates the step /should/ be automated)
145 0. ssh into not-backward, temporarily give the daemon.scripts-security-upd
146 bits by blanching it on system:scripts-security-upd, and run parallel-find.pl
148 1. [ see doc/upgrade.rst ]
150 [ENTER HERE FROM CREATING A NEW REPO]
152 9. Push all of your changes in a public place, and encourage others
153 to test, using --srv-path and a full path.
155 [ XXX: doc/deploy.rst ]
156 GET APPROVAL BEFORE PROCEEDING ANY FURTHER;
157 THIS IS PUSHING THE CHANGES TO THE PUBLIC
159 NOTE: The following commands are to be run on not-backward.mit.edu.
160 You'll need to add daemon.scripts-security-upd to
161 scripts-security-upd to get bits to do this. Make sure you remove
162 these bits when you're done.
164 10. Run `wizard research appname`
165 which uses Git commands to check how many
166 working copies apply the change cleanly, and writes out a logfile
167 with the working copies that don't apply cleanly. It also tells
168 us about "corrupt" working copies, i.e. working copies that
169 have over a certain threshold of changes.
171 11. Run `wizard mass-upgrade appname`, which applies the update to all working
174 12. Run parallel-find.pl to update our inventory
176 [ XXX: doc/upgrade.rst ]
177 * For mass importing into the repository, there are a few extra things:
179 * When mass producing updates, if the patch has changed you will have to
180 do a special procedure for your merge:
182 git checkout pristine
183 # NOTE: Now, the tricky part (this is different from a real update)
184 git symbolic-ref HEAD refs/heads/master
185 # NOTE: Now, we think we're on the master branch, but we have
186 # pristine copy checked out
187 # NOTE: -p0 might need to be twiddled
188 patch -p0 < ../app-1.2.3/app-1.2.3.patch
190 # reconstitute .scripts directory
191 git checkout v1.2.2-scripts -- .scripts
193 # NOTE: Fake the merge
194 git rev-parse pristine > .git/MERGE_HEAD
196 You could also just try your luck with a manual merge using the patch
199 [ XXX: doc/layout.rst ]
200 * The repository for a given application will contain the following files:
202 - The actual application's files, as from the official tarball
204 - A .scripts directory, with the intent of holding Scripts specific files
205 if they become necessary.