Search · Index

III.13 CVS Guidelines

By Joel Aufrecht with input from Jeff Davis, Branimir Dolicki, and Jade Rubick.

OpenACS docs are written by the named authors, and may be edited by OpenACS documentation staff.

All OpenACS code is available anonymously. To get code anonymously, use the parameter -d:pserver:anonymous@cvs.openacs.org:/cvsroot immediately after cvs in a cvs command to check out or export code.

If you are an OpenACS developer, you should check out code so that you or any other developer can commit it. To do this, use the parameter -d:ext:cvs.openacs.org:/cvsroot immediately after cvs in checkout commands. This will create a local checkout directory that uses cvs.openacs.org but does not specify the user. By default, it will use your local account name as the user, so if you are logged in as "foobar" it will try to check out and commit as if you had specified :ext:foobar@cvs.openacs.org:/cvsroot. The advantage of not specifying a user in the checkout command is that other users can work in the directory using their own accounts.

OpenACS.org supports non-anonymous cvs access only over ssh, so you must have CVS_RSH=ssh in your environment. (Typically this is accomplished by putting export CVS_RSH=ssh into ~/.bash_profile.). If your local account name does not match your cvs.openacs.org account name, create a file ~/.ssh/config with an entry like:

Host cvs.openacs.org
User joel

With this setup, you will be asked for your password with each cvs command. To avoid this, set up ssh certificate authentication for your openacs account. (More information)

You may want to set some more default actions for CVS usage. To do so, create the file ~/.cvsrc with the contents:

cvs -z6
cvs -q

-z6 speeds up cvs access over the network quite a bit by enabling compressed connection by default. -q suppresses some verbose output from commands. For example, it makes the output of cvs up much easier to read.

If you are actively developing a non-core package, you should work from the latest core release branch. Currently this is oacs-5-4. This ensures that you are working on top of a stable OpenACS core, but still allows you to commit feature changes to non-core packages. To check out all packages,

cvs -d :ext:cvs.openacs.org:/cvsroot checkout -r oacs-5-4 openacs-4

If you work in the directories created with this command, all of your cvs updates and commits will be confined to the oacs-5-4 branch. Your work will be merged back to HEAD for you with each release.

Because the entire openacs-4 directory is large, you may want to use only acs-core plus some specific modules. To do this, check out core first:

cvs -d:ext:cvs.openacs.org:/cvsroot checkout -r oacs-5-4  acs-core

Then add modules as needed:

cd /var/lib/aolserver/service0/packages
cvs up -d packagename

... where packagename is the name of the package you want. Visit the Package Inventory and Package maintainers and status for a list of available packages and their current state.

If you are actively developing packages in the OpenACS Core, work from the HEAD branch. HEAD is used for active development of the next version of core OpenACS. It may be very buggy; it may not even install correctly. Do not use this branch for development of non-core features unless your work depends on some of the HEAD core work. To check out HEAD, omit the -r tag.

To check out HEAD for development, which requires an OpenACS developer account:

cvs -d:ext:cvs.openacs.org:/cvsroot checkout acs-core

To check out HEAD anonymously:

cvs -d:pserver:anonymous@cvs.openacs.org:/cvsroot checkout acs-core

.LRN consists of a given version openacs core, plus a set of packages. These are collectively packages together to form a distrubution of .LRN. F .LRN 2.0.0 sits on top of OpenACS 5.0.0. .LRN also uses an OpenACS install.xml file during installation; this file is distributed within the dotlrn package and must be moved. To get a development checkout of .LRN in the subdirectory dotlrn:

cvs -d :pserver:anonymous@cvs.openacs.org:/cvsroot checkout -r oacs-5-4 acs-core
mv openacs-4 dotlrn
cd dotlrn/packages
cvs -d :pserver:anonymous@cvs.openacs.org:/cvsroot checkout -r oacs-5-4 dotlrn-all
mv dotlrn/install.xml ..

Once you have a checkout you can use some commands to track what has changed since you checked out your copy. cvs -n update does not change any files, but reports which changes have been updated or locally modified, or are not present in CVS.

To update your files, use cvs update. This will merge changes from the repository with your local files. It has no effect on the cvs.openacs.org repository.

All OpenACS code resides within a single CVS module, openacs-4. (The openacs-4 directory contains code for all versions of OpenACS 4 and later, and .LRN 1 and later.) Checking out this module retrieves all openacs code of any type. For convenience, subsets of openacs-4 are repackaged as smaller modules.

acs-core contains only critical common packages. It does not have any user applications, such as forums, bug-tracker, calendar, or ecommerce. These can be added at any time.

The complete list of core packages is:

ref-timezones search

dotlrn-all contains the packages required, in combination with acs-core, to run the .LRN system.

project-manager-all contains the packages required, in combination with acs-core, to run the project-manager package.

Each OpenACS package (i.e., directory in openacs-4/packages/) is also aliased as a module of the same name.

Tags and Branches look similar in commands, but behave differently. A tag is a fixed point on a branch. Check out a tag to get a specific version of OpenACS. Check out a branch to get the most current code for that major-minor version (e.g., 5.0.x or 5.1.x). You can only commit to a branch, not a tag, so check out a branch if you will be working on the code.

  • openacs-x-y-z-final tags mark final releases of OpenACS. This tag is applied to the acs-core files for an OpenACS core release, and to the latest released versions of all other packages at the time of release. Example: openacs-5-0-4-final.

  • dotlrn-x-y-z-final tags mark final releases of .LRN. These tags apply only to .LRN packages. Example: dotlrn-2-0-1-final

  • packagename-x-y-z-final tags apply to releases of individual packages. For example, calendar-2-0-0-final is a tag that will retrieve only the files in the calendar 2.0.0 release. It applies only to the calendar package. All non-core, non-dotlrn packages should have a tag of this style, based on the package name. Many packages have not been re-released since the new naming convention was adopted and so don't have a tag of this type.

  • openacs-x-y-compat tags point to the most recent released version of OpenACS X.Y. It is similar to openacs-x-y-z-compat, except that it will always get the most recent dot-release of Core and the most recent compatible, released version of all other packages. All of the other tag styles should be static, but -compat tags may change over time. If you want version 5.0.4 exactly, use the openacs-5-0-4-final tag. If you want the best newest released code in the 5.0.x release series and you want to upgrade within 5.0.x later, use the compat tag.

    For example, if you check out the entire tree with -r openacs-5-0-compat, you might get version 5.0.4 of each OpenACS core package, version 2.0.1 of calendar, version 2.0.3 of each .LRN package, etc. If you update the checkout two months later, you might get version 5.0.5 of all OpenACS core packages and version 2.1 of calendar.

  • oacs-x-y is a branch, , not a tag. All core packages in the 5.0 release series (5.0.0, 5.0.1, 5.0.2, etc) are also on the oacs-5-0 branch. Similarly, OpenACS core packages for 5.1.0 are on the oacs-5-1 branch.

    These branches are used for two purposes. OpenACS Core packages on these branches are being tidied up for release. Only bug fixes, not new features, should be added to core packages on release branches. For all other packages, release branches are the recommended location for development. For example, if you are working on calendar, which is compatible with openacs 5.0 but not 5.1, work on the oacs-5-0 branch.

  • HEAD is a branch used for development of core packages.

There are three main ways to contribute code to OpenACS:

  1. To contribute a small fix, if you do not have a developer account, submit a patch.

  2. If you are making many changes, or would like to become a direct contributor, send mail to the Core Team asking for commit rights. You can then commit code directly to the repository:

    1. Use one of the checkout methods described above to get files to your system. This takes the place of steps 1 and 2 in the section called "Installation Option 2: Install from tarball". Continue setting up the site as described there.

    2. Fix bugs and add features.

    3. Commit that file (or files):

      cvs commit -m "what I did and why" filename

      Because this occurs in your personal checkout and not an anonymous one, this commit automagically moves back upstream to the Mother Ship repository at cvs.openacs.org. The names of the changed files, and your comments, are sent to a mailing list for OpenACS developers. A Core Team developer may review or roll back your changes if necessary.

    4. Confirm via the OpenACS CVS browser that your changes are where you intended them to be.

  3. Add a new package. Contact the Core Team to get approval and to get a module alias created.

    1. Check out acs-core on the HEAD branch. (Weird things happen if you add files to a branch but not to HEAD):

      cd /tmp
      cvs -d:ext:cvs.openacs.org:/cvsroot checkout acs-core

      Copy your package directory from your working directory to this directory. Make sure not to copy any CVS directories.

      cp -r /var/lib/aolserver/service0/packages/newpackage /tmp/openacs-4/packages

      Import the package into the cvs.openacs.org cvs repository:

      cd /tmp/openacs-4/packages/newpackage
      cvs import -m "Initial import of newpackage" openacs-4/packages/newpackage mynamenewpackage-0-1d
    2. Add the new package to the modules file. (An administrator has to do this step.) On any machine, in a temporary directory:

      cvs -d :ext:cvs.openacs.org:/cvsroot co CVSROOT
      cd CVSROOT
      emacs modules

      Add a line of the form:

      photo-album-portlet openacs-4/packages/photo-album-portlet

      Commit the change:

      cvs commit -m "added alias for package newpackage" modules

      This should print something like:


    3. Although you should add your package on HEAD, you should do package development on the latest release branch that your code is compatible with. So, after completing the import, you may want to branch your package:

      cd /var/lib/aolserver/service0/packages/newpackage
      cvs tag -b oacs-5-1
    4. See the section called "How to package and release an OpenACS Package"


    Some packages are already in cvs at openacs-4/contrib/packages. Starting with OpenACS 5.1, we have a Maturity mechanism in the APM which makes the contrib directory un-necessary. If you are working on a contrib package, you should move it to /packages. This must be done by an OpenACS administrator. On cvs.openacs.org:

    1. cp -r /cvsroot/openacs-4/contrib/packages/package0 /cvsroot/openacs-4/packages
    2. Update the modules file as described above.

    3. Remove the directory from cvs in the old location using cvs rm. One approach for file in `find | grep -v CVS`; do rm $file; cvs remove $file; done

CVS commit procedures are governed by TIP (Technical Improvement Proposal) #61: Guidelines for CVS committers

  1. Which branch?

    1. For core packages, new features should always be committed on HEAD, not to release branches.

    2. For core packages, bug fixes should be committed on the current release branch whenever applicable.

    3. For non-core packages, developers should work on a checkout of the release branch of the lastest release. For example, if OpenACS 5.1.0 is released, developers should work on the oacs-5-1 branch. When oacs-5-2 is branched, developers should continue working on oacs-5-1 until OpenACS 5.2.0 is actually released.

      Reason: First, this ensures that developers are working against stable core code. Second, it ensures that new package releases are available to OpenACS users immediately.

    4. The current release branch is merged back to HEAD after each dot release.

  2. New packages should be created in the /packages directory and the maturity flag in the .info file should be zero. This is a change from previous policy, where new packages went to /contrib/packages)

  3. Code

    1. Only GPL code and material should be committed to the OpenACS CVS repository (cvs.openacs.org)

    2. Do not mix formatting changes with code changes. Instead, make a formatting-only change which does not affect the logic, and say so in the commit comment. Then, make the logic change in a separate commit. Reason: This makes auditing and merging code much easier.

    3. Database upgrade scripts should only span one release increment, and should follow Naming Database Upgrade Scripts .

      Reason: If an upgrade script ends with the final release number, then if a problem is found in a release candidate it cannot be addressed with another upgrade script. E.g., the last planned upgrade script for a package previously in dev 1 would be upgrade-2.0.0d1-2.0.0b1.sql, not upgrade-2.0.0d1-2.0.0.sql. Note that using rc1 instead of b1 would be nice, because that's the convention with release codes in cvs, but the package manager doesn't support rc tags.

    4. Database upgrade scripts should never go to the release version, e.g., should always have a letter suffix such as d1 or b1.

    5. CVS commit messages should be intelligible in the context of Changelogs. They should not refer to the files or versions.

    6. CVS commit messages and code comments should refer to bug, tip, or patch number if appropriate, in the format "resolves bug 11", "resolves bugs 11, resolves bug 22". "implements tip 42", "implements tip 42, implements tip 50", "applies patch 456 by User Name", "applies patch 456 by User Name, applies patch 523 by ...".

  4. When to TIP

    1. A TIP is a Techical Improvement Proposal ( more information ). A proposed change must be approved by TIP if:

      1. It changes the core data model, or

      2. It will change the behavior of any core package in a way that affects existing code (typically, by changing public API), or

      3. It is a non-backwards-compatible change to any core or standard package.

    2. A proposed change need not be TIPped if:

      1. it adds a new function to a core package in a way that:

        1. does not change the backwards-compatibility of public API functions.

        2. does not change the data model

        3. has no negative impact on performance

      2. it changes private API, or

      3. it is a change to a non-core, non-standard package

  5. Tags

    1. When a package is released in final form, the developer shall tag it "packagename-x-y-z-final" and "openacs-x-y-compat". x-y should correspond to the current branch. If the package is compatible with several different core versions, several compat tags should be applied.

      Reason 1: The packagename tag is a permanent, static tag that allows for future comparison. The compat tag is a floating tag which is used by the repository generator to determine the most recent released version of each package for each core version. This allows package developers to publish their releases to all users of automatic upgrade without any intervention from the OpenACS release team.Reason 2: The compat tags allows CVS users to identify packages which have been released since the last core release.Reason 3: The compat tag or something similar is required to make Rule 6 possible.

    2. When OpenACS core is released, the openacs-x-y-z-final tag shall be applied to all compat packages.

      Reason: This allows OpenACS developers who are creating extensively customized sites to branch from a tag which is stable, corresponds to released code instead of development code, and applies to all packages. This tag can be used to fork packages as needed, and provides a common ancestor between the fork and the OpenACS code so that patches can be generated.

    For example, adding a new API function wouldn't require a TIP. Changing an existing API function by adding an optional new flag which defaults to no-effect wouldn't require a TIP. Added a new mandatory flag to an existing function would require a TIP.

Informal guidelines which may be obsolete in places and should be reviewed:

  • Before committing to cvs you must submit a bug report and patch to the OpenACS bug tracker . The only exceptions to this rule are for package maintainers committing in a package they are maintaining and for members of the core team.

  • If you are committing a bug fix you need to coordinate with the package maintainer. If you are a maintainer then coordinate with any fellow maintainers.

  • If you are to commit a new feature, an architecture change, or a refactoring, you must coordinate with the OpenACS core team first. Also, such changes should have a discussion in the forums to allow for feedback from the whole community.

  • If you are changing the data model you *must* provide an upgrade script and bump up the version number of the package.

  • Consider any upgradability ramifications of your change. Avoid changing the contract and behaviour of Tcl procedures. If you want to build a new and clean API consider deprecating the old proc and making it invoke the new one.

  • Never rush to commit something. Before committing double check with cvs diff what exactly you are committing.

  • Always accompany a commit with a brief but informative comment. If your commit is related to bug number N and/or patch number P, indicate this in the commit comment by including "bug N" and/or "patch P". This allows us to link bugs and patches in the Bug Tracker with changes to the source code. For example suppose you are committing a patch that closes a missing HTML tag, then an appropriate comment could be "Fixing bug 321 by applying patch 134. Added missing h3 HTML close tag".

  • Commit one cohesive bug fix or feature change at a time. Don't put a bunch of unrelated changes into one commit.

  • Before you throw out or change a piece of code that you don't fully understand, use cvs annotate and cvs log on the file to see who wrote the code and why. Consider contacting the author.

  • Test your change before committing. Use the OpenACS package acs-automated-testing to test Tcl procedures and the tool Tclwebtest to test pages

  • Keep code simple, adhere to conventions, and use comments liberally.

  • In general, treat the code with respect, at the same time, never stop questioning what you see. The code can always be improved, just make sure you change the code in a careful and systematic fashion.