net_sccs will be broken into two parts. The first being back end processes that will run under unix, and be interfaced to though http. The second are front end processes which will initially be implemented as command line utilities. The use of http, and the design requirement of keeping the communications layer as simple as possible should not preclude the implementation of GUI based tools that are capable of interfacing to the source control system.
--------------
| command line |
| utilities |
--------------
|
|
------------
| httpd |
| [ www ] |
| [ server ] |
------------
|
|
-----------
| cgi |
| processes |
-----------
|
|
-----------
| interface |
| to |
| sccs |
-----------
|
There are four layers to this design. The first is the front end command line tools. These tools communicate with a web server via http. The web server provides a convienient method for waiting for requests, as well as an interface to the SCCS commands though CGI. The CGI processes will do the acutal work of communicating with sccs -- getting files, checking them out, and checking them in. SCCS will provide the actual archival, and retreival for the source files. |
One known limitation of this system will be that text files will be the only type of data the system is capable of storing. This limitation is inherited from SCCS, which itself cannot handle non-text files.