I don't hate ClearCase. I don't know enough about it to hate it. But I
have a professional SCM friend who swears by it, so I need to withhold
judgement.

I also agree with Matt that you need to look at what your challenge is.
Know what I do at home? For most uses, I make a RCS subdirectory and ci/co
on the tip to get roll back and comments. Period.

If I needed to do something where directory structures or code might be
critical or refactored, or where I had aspirations of code reuse, then I'd
go to svn or learn mercurial/git. Also, p4 has a free (beer) license of
OSS projects.

But, in a professional setting where I was asked what to do, I would make
damn sure everyone knew my opinion of svn's limitations before I put my
fingerprints on it. Too easy to catch the blame when it comes up short.

On Wed, October 24, 2007 5:32 pm, MattyJ wrote:
> I echo what Lan says for the most part, especially the part about doing a
> critical analysis. Your organization might have different needs (all
> Microsoft? All *nix?), philosophies, needs for process built in, etc.
>
> The places I have worked have mostly been early on in their implementation
> of any type of development process, so low-weight/agile tools were a must.
> Perforce has won out twice for me. As Lan said, it's fast, it's
> extensible, hackable and pretty much does exactly one thing and does it
> very well. For source control tools that cost money, it's relatively cheap
> (about $800 per seat at present, I believe.)
>
> I hate to make Lan bristle, but I worked at a place where ClearCase was
> forced upon us. After some animosity I grew to like it to an extent. We
> had six different offices scattered throughout the globe that contributed
> to our code base and ClearCase excelled and managing all the disperate
> pieces. It's also pretty much bulletproof when it comes to availability,
> quality, etc., and it scales to a massive size. If I was hired by Motorola
> to unify their source control/CM functions, I'd seriously consider
> ClearCase (assuming cost is not a factor, it is very pricey, thousands of
> dollars per seat.) It has some drawbacks but as I said, if you have 5000
> developers scattered across the globe ClearCase is the gold standard in
> enterprise source control management.
>
> And to echo what Lan said earlier and to bring the conversation full
> circle, I have nothing against Svn, under the right circumstances. I use
> it at home for personal things, but I have no need for branching, merging
> or even labeling. It suits me well. I have a buddy who owns a small
> consultancy firm (five people) and they use a hosted Svn implementation
> and are extremely happy with it. But again, it's more of a set of
> individual worspaces where they don't need to branch or merge or share
> anything. I think Svn is a solid piece of software in and of itself that
> has the *potential* to be enterprise-ready. It's just the lack of a few
> good features that medium to large shops would need (even if they don't
> realize it yet) that would keep me from implementing it in a bigger
> business. Lastly, alluding again to Lan's comments, as with any tool there
> is a learning curve and certain admin overhead. The non 'out of the box'
> nature of Svn, Cvs, git, etc. is a turnoff for a lot of people.
>
> One final note, and people sometimes think I'm nuts when I say this, being
> the OSS advocate that I am, but I don't use open source software because
> it's free (as in beer or as in freedom), I use it because some of it is
> better. I use a certain amount of commercial software at home because it's
> simply better. I'd like to be 100% OSS all the time, but sometimes I just
> need my damn iPod to work correctly. Or my wife just needs her web pages
> to display. She couldn't care less that Microsoft is evil, and that the
> flash player for Linux is prone to crashing, but you're being a good
> person by not running windows. Sometimes things just need to work and I'm
> willing to pay for that.
>
> So in the long run, if Svn or anything else OSS was better than Perforce
> for my professional situation I'd (try to) change to it in a second. You
> have to ask yourself how much you're willing to sacrifice in time/effort
> vs monetary cost of these things.
>
>
> -Matt
>
>
> On Wed, 24 Oct 2007 13:52:41 -0700, Lan Barnes <[EMAIL PROTECTED]> wrote:
>
>> I'd recommend doing a formal tool search. Get a project manager who
>> knows
>> how to do a tool search and a professional SCM analyst with broad
>> experience. Make a list of candidates and by all means, put svn and
>> other
>> OSS tools on the list. Derive user needs and requirements. Get the input
>> of all stakeholders and whatever regulatory specialists you have. Expand
>> the horizons to include issue traching, perhaps project management
>> tools,
>> etc.
>>
>> There is a wide number of commercial and OSS tools in the arena with
>> different design philosophies. ClearCase and Dimensions have built in
>> process (with some flexibility), which is great if you need that
>> process,
>> not so great if it chafes. Perforce allows you to script damn near
>> anything, but that cost and time are on your head, and you have to
>> support
>> it, too.
>>
>> Obviously budget is in the mix, and don't neglect the cost of
>> customization, internal training, support contracts, consultants,
>> external
>> trainers, and maintenance.
>>
>> If your needs are simple and your process is unsophisticated, maybe svn
>> would work just fine. But we all know that OSS isn't "free." Not if your
>> time has any value.
>>
>> I don't think that there is one best of breed. I'm pretty happy with p4
>> at
>> work, but it's easy to abuse and become a time sink as well. Our present
>> system at work is IMO so deficient in process discipline as to be a
>> negative influence on good SCM practices. It's not important how we got
>> there, we're just there (and no sign of it getting better). And it's not
>> p4's fault at all.
>>
>> Finally, to do good SCM needs the enlightened commitment of upper
>> management. And the real enemy of good SCM is middle management,
>> especially development. Actual developers, once they've worked with good
>> SCM, want it badly and are real champions of SCM. But middle managers
>> see
>> it as a threat to their making schedule (it actually helps) and always
>> want to control it so they can throw it overboard during crunch time,
>> which is when they need control most.
>>
>> Show me a shop where SCM is under Development and/or assigned as a 50%
>> job
>> to one developer on each team, and I'll show you a fsck'd up place.
>>
>> Others may weigh in with different opinions. I don't pretend to know
>> everything about the field, and am forming new opinions every day. But I
>> have been doing it for a while, and I've gotten to learn from some
>> really
>> sharp people.
>>
>> BTW, I'm BCC'ing a few geek friends on this thread who aren't
>> necessarily
>> dialed into Kplug (like Matt). Maybe some of them will have input for me
>> to pass along.
>>
>> On Wed, October 24, 2007 12:06 pm, Bob La Quey wrote:
>>> So what would you, Matt or Lan, recommend for an "Enterprise Level"
>>> SCM?
>> And why?
>>>
>>> What does such a thing cost?
>>>
>>> BobLQ
>>>
>>>
>>> --
>>> [email protected]
>>> http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
>>>
>>
>>
>
>
>


-- 
Lan Barnes

SCM Analyst              Linux Guy
Tcl/Tk Enthusiast        Biodiesel Brewer


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to