Using ClearCase with Fedora

Question

My goal is to use ClearCase (7.0.1+) on Fedora (12+, 64bit) in a company where most of the developers are blissfully using the standard CC/Windows development environment. My most recent experience is with git on Fedora 12 and I find that to be a very comfortable and productive programming environment. Two solutions to this problem spring to mind:

1. Assume it works and fix the breakage: install ClearCase which will attempt to compile support into the kernel. This has been done for some Unix derivatives so, while daunting (I tried this unsuccessfully six months or so ago), there is some guidance that can be googled. As near as I can tell it is a large piece of work.

2. Design and implement a bridge: use cvs/subversion/git/etc. (preferably git) to act as a two-way conduit between actual code development and interactions with the corporate source code repository. This would be a partial bridge (using ClearCase 7.0.1 running on Windows), not a full repository copy in the DVCS sense. Again, some work on bridging with git/CC can be googled, but it is incomplete and often conflicting.

Suggestions for other solutions are welcome but my basic question is: which of these solutions seems to have the best chance of success? Bonus points for suggested approaches for either solution.

Answer 1

It's been a very long time since I last used ClearCase (and even then it was not as an administrator) but I assume that the needed kernel support is for its funky network filesystem for the version-controlled views?

I used a ClearCase feature called snapshot views which basically copied a view into an ordinary filesystem folder, so it could be used detached from the network, without the funky version-controlled network filesystem. Maybe it's possible to installed ClearCase without the kernel support and just use snapshot views? This would get you roughly the same functionality you would get from implementing a bridge to a different VCS.

Answer 2

You might want to have a look at the ClearCase Remote Client (provided your CC repository can be accessed via http). This is a standalone client that isn't any worse than the regular CC client (truth be told, I don't think there is much that is) and it doesn't do anything fancy with own filesystems etc. All it does is create a snapshot checkout on your HDD like a regular version control client.

Answer 3

ClearCase and 64-bit architectures: you will find all the support compatibility-matrix from this article.

From what I can see, neither CC7.0.x or CC7.1.x are supported on Fedora. Period.
You could try to install it anyway as the snapshot view would most likely work (only the MVFS kernel-dependent filesystem would not be installed at all)
Even with snapshot view, it would not be very useful, since all CC operations (checkouts/checkings, compare with previous version, ... need a view able to query the VOB server -- through the albd Atria Location Broker service)
You would have a snapshot view with nothing (no Vob) to query.

---

That being said, for similar "off-line" or "off-site" development, I go with a DVCS (Git in my case), which allows me to:

• commit and branch when and how I want on my laptop at home
• push to my Git repo at work (from a git bundle), considering my Git repo is directly within my ClearCase (snapshot) view.

If the time spend outside ClearCase (in another repo on another computer) is not too long, it works great. The commits you re-import in ClearCase are not too far from the initial code of your ClearCase view, and form a logical new "baseline".