summaryrefslogtreecommitdiff
path: root/docs/historical/ProjectDetailedDesign.trac
blob: e29525107fac0817869b8bb00afe9420e0d259c2 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
This is very much a draft/brainstorm right now. It should be made prettier and thought about in more detail later, but it at least gives some idea of the direction we're headed right now. 
----
 * Two distinct tools; should not be coupled (can work independently):
   * Versioning tool
     * Command line tool; let's call it "samigrate"
     * Organizes old migration scripts into repositories
     * Runs groups of migration scripts on a database, updating it to a specified version/latest version
     * Helps run various tests
     * usage
       * "samigrate create PATH": Create project migration-script repository
         * We shouldn't have to enter the path for every other command. Use a hidden file
         * (This means we can't move the repository after it's created. Oh well)
       * "samigrate add SCRIPT [VERSION]": Add script to this project's repository; latest version
         * If a .sql script: how to determine engine, operation (up/down)? Options:
           * specify at the command line: "samigrate add SCRIPT UP_OR_DOWN ENGINE"
           * naming convention: SCRIPT is named something like NAME.postgres.up.sql
       * "samigrate upgrade CONNECTION_STRING [VERSION] [SCRIPT...]": connect to the specified database and upgrade (or downgrade) it to the specified version (default latest)
         * If SCRIPT... specified: act like these scripts are in the repository (useful for testing?)
       * "samigrate dump CONNECTION_STRING [VERSION] [SCRIPT...]": like update, but sends all sql to stdout instead of the db
       * (Later: some more commands, to be used for script testing tools)
   * Alchemy API extensions for altering schema
     * Operations here are DB-independent
     * Each database modification is a script that may use this API
     * Can handwrite SQL for all databases or a single database
     * upgrade()/downgrade() functions: need only one file for both operations
       * sql scripts reqire either (2 files, *.up.sql;*.down.sql) or (don't use downgrade)
     * usage
       * "python NAME.py ENGINE up": upgrade sql > stdout
       * "python NAME.py ENGINE down": downgrade sql > stdout