Community technical support mailing list was retired 2010 and replaced with a professional technical support team. For assistance please contact: Pre-sales Technical support via email to sales@march-hare.com.
Hi Aditya, Comments inline... Aditya Gandhi wrote: > HI all, > I am planning to use CVS for my code repository and would like to be > able to associate all the code commits to some Change Requests (tracker > items). > My tracker does not expose any API therefore interacting with it and > adding metadata at that level is not a possibility. > However, can we do something in CVS to address this- > I had some thoughts on the same and am looking at the group to help me > refine or explore some better alternatives we have a similar product; we can interface with the API, but we have chosen not to. > 1. When ever somebody checks in a file associated with a CR then tag > that file with this CR. > If a file is checked in multiple times one might just have info > about the last checkin (or can this be avoided.) > 2. Building on top of this on commit one can scan thru the comments for > the CR number and in the loginfo trigger ask for the file to be > automatically tagged by picking up the CR number from comments (Will > need to take care of self locks) this is similar to what we do, but instead of using a tag, we simply add the number in a specified format to the comment. We then use viewcvs with it's commit database to search through all the commits for one (or many) that match the specified text. This way, you will find multiple commits on a single file. We also have a hyperlink format that launches our bug tracking software with the specified ticket numer when it is clicked in the browser (viewcvs). > 3. Create a macro in WinCVS that does the following - When ever such a > report is needed just query all the items and scan thru the logs to look > for a match for all the checkin comments that had the given CR number in > it and then based on that show the parsed output. > Will this be a big performance problem if the sample set looks > something like this - 500 files with an average 20 revisions per file > as mentioned above, I would use viewcvs. It will build a mysql database with all your commits and files and authors, etc, etc. This database then has a full text index for fast searching. We are running it on a windows 2000 server, and we've had no trouble. I'm not sure of the size of the database, but it runs very fast, and we have a fairly large repository. good luck, -kz