Today's Lecture

• Versioning
• A specific versioning system: RCS

Version Control (1 of 3)

• Things change…
  • new requirements lead to…
  • new or changed designs that lead to…
  • new or changed source code, etc.
  • or bugs are discovered and must be fixed
• …software engineers need to keep track of all these changes

Version Control (2 of 3)

• Versioning
  • Keeping track of the changes to a file from one editing session to the next
  • Computer Science has developed algorithms that can automatically detect the changes to a file
    • especially text files
  • One such algorithm is encapsulated in a Unix tool called diff

int main() {
    printf("hello world")
}

int main() {
    printf("HELLO WORLD!")
}

$ diff world1.c world2.c
2c2
<  printf("hello world");
---
>  printf("HELLO WORLD!");

$ diff world2.c world1.c
2c2
<  printf("HELLO WORLD!");
---
>  printf("hello world");

• More about Diff
  • Command Line: diff file1 file2
    • "show me how to change file1 into file2"
  • Output: display the "ed" commands needed to change file1 into file2
    • "2c2" in the example above meant change line 2 in file1 into line 2 of file2
  • See the diff and ed man pages for complete details

Version Control (3 of 3)

• We need to track all the changes associated with a file Why?
  • Rollback: Changes are not always an improvement. Often, you need to return to a prior version
  • Experimentation: Lets software engineers explore “what if” scenarios
  • Internationalization: Different versions for different languages
  • Historical Record: A customer reports a bug on version 1.1, but you are at version 2.3. You need a way to recreate version 1.1 to check out the bug

Tracking Changes

Version Graphs

• All changes to a single file are tracked via a graph
  • nodes are distinct versions
  • edges denote that the target node was derived from the source node
• There are three possible configurations
  • Extension: a single version derived from a single version
  • Split: multiple versions derived from a single one
  • Merge: a single version derived from multiple versions; merge typically done “by hand”

Version Control Files

• Version graphs are stored in version control files
• Space saving techniques can be employed
  • Forward Deltas
    • The original version of a file is stored, all subsequent versions are stored as sets of changes or “deltas”
  • Backward Deltas
    • The most recent version of a file is stored, all previous versions are stored as deltas

Version Control Systems

• Version control systems let you manage the versions contained within a version control file
• You can add a new version, check out a copy of a previous version, remove a version, etc.
• For example: If I ask for version 2.2 of a file that is managed by a forward delta system then it starts with the original file and applies all of the deltas that lead to version 2.2 in the version graph

RCS: Revision Control System

• One “famous” version control system is RCS
• RCS is a backward delta versioning system
  • It stores the complete text for the most recent version of the file
  • And derives old versions by applying deltas to the recent version
• Commonly used RCS commands
  • ci — check in
    • Used to check changes into a version control file; can also create an initial version control file
  • co — check out
    • Used to check out a specific version of a file

RCS Locks

• RCS has a notion of locks
  • If you check out a file, with a lock, no other person can check out that file for editing (they can still read it, however)
  • This prevents multiple people from changing the same file at once
• Not all versioning systems make use of this concept
  • Subversion and CVS, for instance, allow a file to be edited by multiple people
    • If a conflict develops between the different copies, they will ask a developer to perform a merge that resolves the conflict

RCS Version Control File

• The version control file for RCS has the same name as the file under revision control but adds a ",v" suffix
• You can think of this file as a tar file
  • A single file with multiple files stored within
  • However, in this case the contained files are just different versions of the same file and all but one are stored as “deltas”

RCS Example

rw------- CommentBook

$ ci CommentBook
initial revision: 1.1
r-------- CommentBook,v

$ co CommentBook
r-------- CommentBook
r-------- CommentBook,v

r-------- CommentBook
r-------- CommentBook,v

$ rm CommentBook
r-------- CommentBook,v

$ co CommentBook
r-------- CommentBook
r-------- CommentBook,v

$ co -l CommentBook
rw------- CommentBook
r-------- CommentBook,v

$ ci CommentBook
new revision: 1.2
r-------- CommentBook,v

RCS Keywords

• RCS will look for certain keywords in a file and will substitute values for them during check-out
  • Think of these words as variables and RCS provides the definition for each one when a file is checked out
  • Keywords are delimited with two dollar signs
    • Example:if $Author$ appears in a text file stored in RCS then when checked out, RCS replaces the above string with: $Author: kena $ or more generically with $Author: <username> $
    • Run man co for a list of all the keywords defined by RCS

RCS Version Numbering

• An RCS version number has a branch number and a version number. The version number is always the last digit

Other RCS Tools

Coming Up Next

• Lecture 13: Configuration Management
• Lecture 14: Subversion
• Quiz 5: Available after this lecture. Don't forget to take it!!!
• Homework 6 Assigned