Boost logo

Boost :

From: Fernando Cacciola (fernando_cacciola_at_[hidden])
Date: 2006-09-16 11:10:34

Hi People,

A few months ago I read someone (can't remember who, sorry) recommending the
use of 2 working copies (one for the head and one for the release candidate)
instead of swithicng back and forth one and the other.

Asumming that is indeed a good idea, can anyone confirm the following
release procedures?

"boost/HEAD" is the folder containing the working copy that corresponds to
the main trunk
"boost/RC_X_Y_Z" is the folder containg the working copy for the current
release candidate

(0) Install the working copies

  cd boost/HEAD
  cvs checkout boost

  cd boost/RC_X_Y_Z
  cvs checkout -r RC_X_Y_Z boost

(1) Commit local fixes in the head the to main branch (called HEAD in CVS)

  cd boost/HEAD
  cvs commit -m "Committed fixes in trunk" buggycode.hpp

(2) Update the release candidate branch
    cd boost/RC_X_Y_Z
    cvs update (no need to specify the branch, as in -r RC_X_Y_Z, in this
case )

(3) Merge changes in a trunk since previous merge to branch
    ** still in boost/RC_X_Y_Z **
    cvs update -jmerged_to_RC_X_Y_Z -jHEAD buggycode.hpp
    --> RCS file: /cvsroot/boost/.../buggycode.hpp,v
    --> retrieving revision 1.4
    --> retrieving revision 1.6
    --> Merging differences between 1.4 and 1.6 into buggycode.hpp

(4) Commit merged branch
    ** still in boost/RC_X_Y_Z **
    cvs commit -m "Merged fix for problem xyz from trunk to branch"

(5) Go back to main trunk

    cd boost/HEAD
    cvs update (no need to physically switch back to trunk adding "-A" in
this case)

(6) Move tag to a new merged point
    ** now in boost/HEAD **
    cvs tag -F -c merged_to_RC_X_Y_Z buggycode.hpp

(7) Repeat as needed


Fernando Cacciola

Boost list run by bdawes at, gregod at, cpdaniel at, john at