Is it good to use CVS to manage(version control) Cadence lib files? cheers, Ming
This product integrates CVS management into the DF2 environment: http://www.cliosoft.com/products/sos_viadf2.html I don't have any personal experience with it but some people at work swear by it.
For heavy stuff, you can use perforce, clearcase, or synchronicity. That will require an expert to be assigned to it s maintenance and support. For lighter, you can use cadence's own versionsync or CVS. I know of at least 3 PDK publishers (foundries) who offer in their PDK a CVS integration in df2. They probably all lack the most elementary features, can all can bring libmanager to it s knees, and confuse your designers. But the heavier stuff can also confuse them on multiple sites. So I have no particular recommendation, make your pick...
I've used CVS on reference libraries ( or PDKs) to keep track of my edits on source and binary (cdba) files without any problems. This is not the usual mode that a user would be interested in though. I've also recently heard of a company that wrote some skill wrappers to interface dfII into a CVS database. They were very happy with it. Mark
if you are the only one working on the libraries it works just fine. I do not think there is a gdm layer already written for cvs. I normally use CVS to manage my pcell libraries. I use CVS to manage the source code, but also enter the compiled pcells into a cvs managed module. The problem is if someone else checks in a file to the repository, then when you check your file in you'll get a conflict which looks to the design framework simply as a corrupted file. The only way around this i have found is to remove the cdb files from the directories, do a cvs update, then compile the pcells again. -jim
Short answer is No. The problem of version control is one in which you need a very good idea of who will have write access to the data. CVS/dfII is not good with multiple people having write permission on one version controlled entity. Several models of version controlling can be set up, each with it's own set of problems and advantages. If a Librarian handles all of the data writes to the master, several good systems ca be built. One system I have used involved writing a "skillOut" function that essentially wrote a skill script that would re-create any layour/schematic. This file (that when run would recreate the cell) was put under CVS (or RCS) control by a master librarian and a "build" script was written that would re-generate the library. With some clever symbolic links, users would always se the latest library, and would submit changes to the librarian ... (then script were written to automate the librarian & build functions ... ) Good luck -- Gerry