bazaar performance with single large project and a comparison with?git / mercurial

[Talden]

>  > > Our developers are a mix of Windows, Cygwin and likely to soon also
>  > > include linux users.  The users are located in three different
>  > > continents with two of the three teams using CVS over a WAN with some
>  > > fairly severe latency, bandwidth and unfortunately reliability issues.
>  >
>  > How does ClearCase work over WANs?
>  > As for ways to rule out ClearCase, have you tried to come up with
>  > a benchmark that measures the server's load?  It may be irrelevant in
>  > some sense, but management may not notice whether it's relevant.
>
>  From what I hear (which may be urban myth but...) ClearCase
>  pragmatically requires dedicated sys admins for any sizeable
>  installations, not to mention the licence costs.  Any cost conscious
>  manager will therefore not want to use ClearCase so to avoid charges to
>  the budget.

This is not a myth.  Conversations with CMs in our new corporate
parent confirm that they'd expect dedicated staff, possibly even one
per site.  ClearCase was a fairly recent standards decision in the new
corporate - I'm yet to be convinced that they looked at and understood
the entire SCM/VCS space before they made their decisions - they have
also set similar standards for most of the rest of the rational
toolbox and are already having issues.  Reduced cost would seem
unlikely to be a strong argument - this is a multi billion-dollar
company and they seem to think that cost is a necessary component of
this process.

I have had some small team experience with ClearCase and found it
painful then - other than UCM nothing much appears to have improved.
I'm ever hopeful that our tiny team can demonstrate that their choice
of process is a major reason for their agility and success and that we
won't be forced into a prescriptive system that stifles our
productivity.

I think that, because these OSS projects aren't commercially driven
that they don't feel the need to market themselves by attacking the
weaknesses of their commercial competitors where it hurts - cost of
operation is more than just licenses, it's also management overhead,
productivity loss and, to some degree a loss of freedom to innovate.
The DVCS's need to explain why they're a stronger choice and convert
the market before the commercial tools with their brand identity move
to adopt the same approaches - if we want free software to win out it
needs to work harder to show that it's not just 'as good' in many
cases but often 'better'.

--
Talden