SVN tutorial n 10 - merge

The svn merge command applies the differences between two sources to a working copy path.

The synopsis is:

 svn merge [-c M[,N...] | -r N:M ...]  <source>[@svn_revision] <target>

or

 svn merge --reintegrate  <source>[@svn_revision] <target>

or

 svn merge <source1>[@svn_revision_N] <source2>[@svn_revision_M] <target>

In all three forms <target> is the working copy path that will receive the differences. If <target> is omitted, the changes are applied to the current working directory, unless the sources have identical basenames that match a file within the current working directory.  In this case, the differences will be applied to that file.

In the first two forms, <source> can be either a URL or a working copy path (in which case its corresponding URL is used). If the peg revision "svn_revision" is not specified, then HEAD (latest svn revision) is assumed. In the third form the same rules apply for <source1>, <source2>, "svn_revision_N", and "svn_revision_M" with the only difference being that if either source is a working copy path, then the peg revisions must be explicitly stated.

Sync and Cherrypick Merges

The first form, when used without either the -c or -r options, is called a “sync” merge and -r 1:svn_revision is implied. This variant is used to merge all eligible changes to a branch from its immediate ancestor branch.

When the first form is used with the -c or -r options, this is called a “cherrypick” merge and is used to merge an explicitly defined set of changes from one branch to another.

Multiple -c and/or -r instances may be specified, and mixing of forward and reverse ranges is allowed— the ranges are internally compacted to their minimum representation before merging begins (which may result in a no-op merge or conflicts that cause the merge to stop before merging all of the requested revisions).

In both variants of the first form, <source> in revision "svn_revision" is compared as it existed between revisions "svn_revision_N" and "svn_revision_M" for each revision range provided.

Reintegrate Merges

The second form is called a “reintegrate merge” and is used to bring changes from a feature branch (<source>) back into the feature branch's immediate ancestor branch (<target>). Reintegrate merges support only this specialized use case and as such have a number of special requirements and limitations that the other two merge forms do not posses.

2-URL Merges

In the third form, called a “2-URL Merge”, the difference between <source1> at revision "svn_revision_N" and <source2> at revision "svn_revision_M" is generated and applied to <target>.
The revisions default to HEAD if omitted.

If Merge Tracking is active, then Subversion will internally track metadata (i.e. the svn:mergeinfo property) about merge operations when the two merge sources are ancestrally related—if
the first source is an ancestor of the second or vice versa—this is guaranteed to be the case when using the first two forms. Subversion will also take preexisting merge metadata on the working copy target into account when determining what revisions to merge and in an effort to avoid repeat merges and needless conflicts it may only merge a subset of the requested ranges.

Merge Tracking can be disabled by using the --ignore-ancestry option.

Unlike svn diff, the merge command takes the ancestry of a file into consideration when performing a merge operation. This is very important when you're merging changes from one branch into another and you've renamed a file on one branch but not the other.

Examples:

Merge a branch back into the trunk (assuming that you have an up-to-date working copy of the trunk):

 svn merge --reintegrate https://mysvnrepo/svn/branches/myproject-branch-1.0

 svn commit -m "Merge myproject-branch-1.0 back into trunk."

To merge changes to a single file:

 svn merge -c 31 ^/trunk/myfile.txt myfile.txt 

Merge between two revision revisions:

 svn merge https://mysvnrepo/svn/branches/myproject-branch-1.0 -r103:HEAD