Pretty much just ClearCase (was Re: ClearCase Vs HG)

Mark A. Flacy mflacy at verizon.net
Sun Apr 22 00:36:35 CDT 2007


On 2007.04.21 20:44, Michael P. Soulier wrote:
> 
> It always amazes me that people refer to ClearCase as supporting  
> multi-site. When I worked for Nortel, we had a proprietary version  
> control system (unfortunately it remained so) that handled multi-site  
> development in ways that ClearCase can still only dream of doing.

Well, as long as you developed in PROTEL or didn't mind that your C  
files ended in .AA15 (or something) or didn't mind that everything was  
in the same directory. 
> Rational's idea of multi-site was to have a stream per region, with  
> some poor schmuck given the job of "integration prime", sitting in  
> diffmerge all day pulling in patches. Nothing like paying a guy $70000  
> a year to run diffmerge.

Nortel uses NEU (Nortel Enhancements to UCM), which provides the  
reasonchecking, changeset management, and integration into the change  
requestdatabase that you were used to from PLS.

In fact, there's now integration into a code inspection tool so that you  
can't deliver into the  main integration stream unless you have a valid  
code inspection record that is in a certain state.

Multi-site is handled via VOB replication, so there is a delay in seeing  
a change delivered to some other site.
UCM is a major improvement over basic ClearCase.  Instead of diffmerge,  
you just deliver the activity (i.e. changeset) into a target stream.   
Your activity can contain multiple versions of the same file.


> I would argue that comparing them to ClearCase is like comparing a   
> hybrid  economy car to a gas-guzzling SUV. They'll both get you there,  
> but one  is  really slow and expensive, and bad for the environment.  
> The SUV has  special  capabililties that people tout, like the ability  
> to go offroad, but  typically  never use. 
> Mind you, SUVs are simple to drive. ClearCase on the command-line on   
> unix gives command-line interfaces a bad name.

What, compared to PLS?  You're using rose-colored glasses with that one.

PLS was an *excellent* tool for PROTEL development.  Not so great for  
other languages.

This is getting very OT for a Mercurial list.  I'll make any further  
responses off the list unless specifically asked otherwise.

And mflacy at verizon.net does not speak for Nortel but only for himself.   
Just to be clear about it.



More information about the Mercurial mailing list