There are times when you receive a third party code drop which you wish to import. The classic method is documented in Tech Note 15 and its reference to working disconnected (Tech Note 2). The techniques mentioned work very well to find new files, deleted files and changed files.

There is sometimes a fly in the ointment to do with keyword expansion. This is things like a CVS code drop containing expanded keywords:

$Id: //depot/robertcowham.com/main/blog/data/scm/p4_handling_keywords.html#1 $

The Perforce equivalent of this might be:

$Id: //depot/robertcowham.com/main/blog/data/scm/p4_handling_keywords.html#1 $

The simple command to find differences is "p4 diff -se". If your local version has Perforce keyword expansion turned on then you will get a load of files spuriously identified as having changed where the only real change is in the keywords.

Thus we want a simple script to run through the diffs and exclude any diffs where only keywords are found (note that this includes where the keyword is embedded, such as in a static variable assignment).

The following simple script is a good base for this. It does the job, and performs pretty well, handling thousands of files in a few minutes. It makes use of unified diff format where changed lines have a prefix in the first character of the output.

# Script to import a set of changed files with existing keywords already expanded
# (either Perforce or CVS).
# Does "diff -se" and processes the output

# Args: current directory to check
  
require 'P4'

p4 = P4.new
p4.tagged
p4.connect

def process_file(p4, f)
  diffs = p4.run_diff("-f", "-du", f)
  real_diffs = Array.new
  diffs.each { |line|
    case line
    when /^====/
    when /^\@\@/ 
    when /^ /
    else
      if line !~ /\$Id|\$DateTime|\$Revision|\$Date|\$Author|\$Name|\$RCSfile|\$Source/
        real_diffs << line
        # puts f, line
      end
    end
  }
  if real_diffs.size > 0
    print "Editing #{f}\n"
    p4.run_edit(f)
  end
end

all_files = p4.run_diff("-se", ARGV[0])
print "Processing #{all_files.size}\n"
i = 0
all_files.each{|f|
  i += 1
  # print"Processing #{i}\r" if i % 10 == 0
  # print"Processing #{f['depotFile']}\n"
  process_file(p4, f)
}

It is pretty easy to run, e.g.:

diff_se.rb ...

The net result will be a list of files checked out (p4 edit) in the default changelist.

Note that one of the big advantages of Perforce branching and merging is that it handles merges neatly when keyword expansion is used between branches (and thus you don't get spurious conflicts).

CVS Imports

If you use the cvs2p4 scripts to import a CVS repository you can end up with a slight problem since the conversion copies the CVS archive files (in RCS format) and Perforce uses them unchanged. The problem comes about because CVS stores the keywords already expanded in the RCS archive. Perforce stores its RCS files with the keywords not expanded, which makes it easier for it to do the merging between branches (without keyword conflicts). While Perforce can handle a CVS archive with the the keywords "pre-expanded", it does lead to spurious merge conflicts. Note that this problem is only really present during the early merges after the CVS import. It will no longer be present as soon as the base file for any merge is fully in Perforce format (i.e. after at least one merge has been done).