Hi Marko: On Mon, 17 Mar 2008, Marko Niinimaki wrote:
> I find nothing very much wrong with CVS Well, to boot you cannot even rename a file without loosing its history. (There exists a kludge by direct manipulation of the CVS repo data, but it does not work well if one tries to pull historical versions again.) Other CVS shortcomings are listed at <http://en.wikipedia.org/wiki/Concurrent_Versions_System#Limitations> or under "Subversion's Features" at <http://subversion.tigris.org/>. > I've seen that SVN has a pretty good integration with a Savannah > type task tool Yup. <https://gna.org/file/tracker-svncommits.jpg?file_id=1604> But CERN's Savannah does not offer bundled SCM support, and is not really integrated with CERN's SCM tool either to offer this kind of functionality. <https://savannah.cern.ch/support/?func=detailitem&item_id=101661> > In practice: (1) someone reports a bug or a feture request (2) the > task is given to X (3) X does the task, submits the changes to SVN, > reports changeset number in the task tool (4) Y studies the code, > accepts for alpha, the trunk->alpha release is done using the task > tool. Personally I never felt a need to integrate the task tool with the SCM tool too tightly. We started by simply maintaining a TODO list inside SCM, and adopted Savannah only later, but we still use it very lightly, basically only to note down longer-term tasks. So it was always SCM that had the definitive say. We'd occasionally refer to the task/support number in the commit log, such as: 2006-04-27 Tibor Simko <[email protected]> * modules/websearch/lib/search_engine.py: In the text MARC output format, do not output multiple times the same tag in case user specified it more than once. (Closes task #3384.) Did you use some particular task tool + SCM tool combination in the past that was very light and transparent? As if: if there's no Emacs interface, then... ;-) Best regards -- Tibor Simko ** CERN Document Server ** <http://cds.cern.ch/>
