| 1 | $Id: readme.txt 46 2006-01-13 18:54:57Z jari $ |
|---|
| 2 | |
|---|
| 3 | svnstat traverses a directory structure (controlled by subversion) and |
|---|
| 4 | calculates developer statistics for all subversion controlled |
|---|
| 5 | entries. The program takes requires at least one argument - the top |
|---|
| 6 | level directory of the directory structure to traverse. The result is |
|---|
| 7 | written to a sub-directory, svnstat, that will be created in the |
|---|
| 8 | current working directory. |
|---|
| 9 | |
|---|
| 10 | To understand what statistics is calculated by svnstat this definition |
|---|
| 11 | is needed: The developer who made the latest change to a line still in |
|---|
| 12 | use in the latest (checked out) revision, is considered as the |
|---|
| 13 | contributor of that line regardless of who actually originally created |
|---|
| 14 | that line. |
|---|
| 15 | |
|---|
| 16 | The statistics calculated is the number of lines contributed from each |
|---|
| 17 | contributing developer in the latest (checked out) revision. Also, the |
|---|
| 18 | number of lines still in use from each revision is calculated. |
|---|
| 19 | |
|---|
| 20 | There are many different types of files and for many file types it |
|---|
| 21 | does not make sense to define lines. Source code, documentation, and |
|---|
| 22 | other human readable files can be treated on line basis whereas |
|---|
| 23 | symbolic links and binary files cannot. svnstat treats non-line based |
|---|
| 24 | files as one-line files, i.e. the developer that made the last change |
|---|
| 25 | to this type of files gets a one-line credit for it. |
|---|
| 26 | |
|---|
| 27 | How should we exclude un-reasonable credit for large line based files? |
|---|
| 28 | We could add a property to that file, tagging it as a file that should |
|---|
| 29 | be treated as an non-line base file. |
|---|
| 30 | |
|---|
| 31 | The current design assumes that the subversion repository to be |
|---|
| 32 | analysed is checked out, maybe one would want to run the analysis |
|---|
| 33 | agains a subversion repositoy directly. |
|---|
| 34 | |
|---|
| 35 | The current flow of the program is. |
|---|
| 36 | |
|---|
| 37 | i) Extract the directory structure starting from the directory given |
|---|
| 38 | at the command line. |
|---|
| 39 | |
|---|
| 40 | ii) Parse the directory structure, removing files and direcotries that |
|---|
| 41 | is not in subversion control. After this step, only valid |
|---|
| 42 | subversion entries will be allowed in the tree. Beware, 'svn blame' |
|---|
| 43 | chokes on directories, and the directories are still a part of the |
|---|
| 44 | tree. |
|---|
| 45 | |
|---|
| 46 | iii) Walk through the directory structure and calculate statistics for |
|---|
| 47 | each entry. |
|---|
| 48 | |
|---|
| 49 | iv) Compile statistics as discussed. This step may be intermingled |
|---|
| 50 | with step iii). |
|---|
| 51 | |
|---|
| 52 | v) Create the html presentation as discussed. |
|---|
