Software Tools Minutes 16 February 1996 9:00AM - 10:30AM ====================== Present: Stu Fuess, Krzysztof Genser, Herb Greenlee, John Hobbs, Alan Jonchkeere, Stan Krzywdzinski, Qizhong Li-Demarteau, Lee Lueking, Laura Paterno, Harrison Prosper(by phone) Scribe: Laura Paterno Krzysztof started off the meeting by discussing PCL-CVS which is a thin EMACS layer over CVS. Basically, it allows you to invoke CVS commmands from EMACS. This includes things like updating, fetching, creating, etc. (as per CMS) information in the CVS repository. Currently we do not have it here. But if we chose to use it we can tailor it for D0. Alan then proceeded to go through his 3 Handouts. They are located in the following directory: D0SFA::USR$ROOT:[JONCKHEERE.TOOLS] and called D0LIBRARY.TXT (1st handout) CODE_MGR.TXT (2nd handout) CODE_STRAW.TXT (3rd handout) We skipped the first handout (dated 20-Oct-1995) as it had already been discussed in a previous meeting and proceeded to the second (dated 15-Feb-1996) titled D0 Code Management - General Comments and Questions. A number of questions arose from this document and I list them below for further thought. 1) How much data should we store for code testing purposes? 2) How do we handle machine dependent libraries? 3) How do we handle changes which affect multiple libraries? What is meant here is that when we release low level code lots of things may use that code. Do we attempt to try to recompile and test all the code that depends on it as part of the release procedures or is that up to the individual czar to take care of? 4) Do we make the library one big OLB and pull OBJ files to create a product or do we have separate modules for each product? From the user's point of view it would be nice to see as few products as possible. 5) How do we define a frame? They have gotten out of control in our current scheme. 6) Should the library release be a push or a pull mechanism? It was also pointed out that a lot of tools are going to be needed to make whatever we chose as invisible to the user as possible. This will require manpower. After discussing the second of Alan's documents we went on to the third document (dated 14-Feb-1996) titled Code Management Strawman. After going through the document it was agreed that the interface to the code management system should get the most focus. Stu raised the point that we should also consider the timing issue involved with development that is going on now.