Having/being a relatively heavy user of Clearcase I would disagree with you. Clearcase is a mature product. It's problem is that it has so many options and possibilities that it is trivial to shoot yourself in the foot. In order to set it up for usage by non-CC gurus, you have to put in quite a lot of thought. But I would agree that it most likely did cause things "generally, to become slower, less intuative, and more difficult.". It takes up many more resources, and once one person did something wrong, the problems tend to compound as other users try to work around the initial problem. This is not a system to be taken lightly.
As you indicated, CVS+bugzilla is usually a much better solution for small to medium size groups.
Is a 300 developers shop considered "medium" or "large"? If the later, I'm afraid we just ran out of group sizes that CC IS good for.
The original setup was CVS+ClearQuest. ClearQuest was very slow. CVS wasn't too fast either, and had serious locks problem (i.e. - people would constantly have locks on directories because they were commiting/tagging). The later problems affected mostly project managers (as the rest would only need a lock on a single directory, and thus were not too much bothered by the problem). In any case, an elaborate system (that was necessary anyways) was put in place to seperate the company's source code into almost independant modules, so that the problem was really affecting very few people.
After the switch to clearcase+clearquest, several things became apparent:
A. Integration didn't work. The theory that CC and CQ can integrate didn't stand up to real life. Things crawled to a standstill the moment that was turned on.
B. ClearCase didn't respond well to being hosted on NetApp. Nobody knew exactly what the problems were, but strange problems would happen all the time (see - 500MB quota story).
C. CC didn't scale. Each server could only run 1000 processes, and as each view requires a server process, there were constant requests from the admin team for people to delete views.
D. While the locking problem indeed went away, the average time it took to checkout a new project, or to perform any other operation, had actually increased rather than decreased.
E. No decent command line support. The beutiful diff tools are only available as GUI. No convinent "cvs diff -u". As we all know, the effeciveness of GUI has a certain cap. It was very annoying to bump against it.
Now, I have not been there for quite some time. Maybe some or all of these problems got better as administrators got better aquented. Maybe the generation that knew how things have been on CVS is slowly leaving, so that people no longer know that they are having a bad system. Maybe they are still cursing under their lips every time they have to do a CC operation. I don't know. I do know that the migration took over a year, with the close help of Rational.
I agree that many of these probelms probably won't affect people working in smaller groups. Then again, these people can probalby be as effective with CVS, since the added functionality CC gives there is less important. To me, this spells "you're damned if you do and you're damned if you don't".
Shachar
-- Shachar Shemesh Open Source integration consultant Home page & resume - http://www.shemesh.biz/
================================================================= To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
