Hello and Good Day.

We are in the process of selecting a version control for our inhouse
development. We will have about 25 developers coding COBOL, JAVA,
shell and perl scripts under HPUX. We are not a complicated shop and
really are not looking to pay much for VC. Today on the system we
migrating from, we use a home grown system I wrote - it's simple, but
it gets the job done. However, that system will not be ported to UX
and so we either need to buy or put together a shareware solution,
whatever our choice, it needs to be able to handle the following list
of items:

Allows conconcurrent development of same source
Manager Approval feature
Enforce standards feature though VC
Atomic promotions with remote compile
Developer driven Rollback
Log / Reporting features
Source difference feature
UX Command line interface
Web browser interface
IDE Interface
Security user levels with encrypted passwords
Can script around code
Requires NO third party database
Runs under HPUX
Support Options - email,phone and list server
Online Documentation
Competitily priced.

And here's the list of products we will soon be researching:

perforce = http://www.perforce.com/
borland starteam = http://www.borland.com/starteam/
borland together = http://www.togethersoft.com/products/
cvs = http://www.cvshome.org/docs/manual/
razor = http://www.visible.com/products/razor/index.html
aim = http://www.innovate.com/products.html
aldon affinit = http://www.aldon.com/affiniti/
marant= http://www.merant.com/Products/ECM/ecmsolutions.asp
webiso = http://webiso.us/index.html
bitkeeper = http://www.bitkeeper.com/

There is a good chance we will go with CVS, not mater what our
findings only becuase it's so damn popular and there is so much
support information online. Not the mention the FREE thing. However,
need to perform due diligence in making sure we select the best
product for the firm.

Any help or information is greatly appreciated. Especially any
information on if and how CVS can meet our (simple) list of
requirements.

Thanks in Advance!!!

In article <ef0a04d7.0404141105.6ad2da21@posting.google.com >,
jason@cyberpine.com wrote:

> We are in the process of selecting a version control for our inhouse
> development. We will have about 25 developers coding COBOL, JAVA,
> shell and perl scripts under HPUX. [...]

Two recent SCM you didn't list are

Subversion http://subversion.tigris.org/
GNU arch http://www.gnu.org/directory/arch.html

The former has recently reached version 1.0, the latter, 1.2.

Subversion is meant as the successor to CVS, without the
limitations of the original (the lack of support for directories
and symlinks, renaming, atomic commits, changesets, undo, etc.)

GNU arch is more powerful than Subversion, but also more
complex.

CVS became very popular because it was more or less the only
libre SCM allowing concurrent access. Now that better
alternatives are available, I expect that most CVS users will
gradually move to either Subversion or GNU arch. My employer (a
Unix ISV) plans to do so in the short term.

--
André Majorel <URL:http://www.teaser.fr/~amajorel/>
"Finally I am becoming stupider no more." -- Paul Erdös' epitaph

Andre Majorel <amajorel@teezer.fr> wrote in message >
> Subversion is meant as the successor to CVS, without the
> limitations of the original (the lack of support for directories
> and symlinks, renaming, atomic commits, changesets, undo, etc.)
>
> GNU arch is more powerful than Subversion, but also more
> complex.
>
> CVS became very popular because it was more or less the only
> libre SCM allowing concurrent access. Now that better
> alternatives are available, I expect that most CVS users will
> gradually move to either Subversion or GNU arch. My employer (a
> Unix ISV) plans to do so in the short term.

Thanks for your response, very helpful right now.

Some newbie questions

Are those limitation still true for CVS?

Lets say I wanted to encapsulate the promotion process in a script and
test that sources for certain code standards and reject the promotion
if that naming convention or actual code standard fails? Possible with
all three products? Possible in likely the same ways?

With such a large install base for CVS, would it not be safer (from a
support point of view) to stay with the croud for now. Reason I ask is
that management is hesitant to go freeware unless we can prove that
the list servers and online doc are as good as or better than phone
support you get with a commercial product.

All three are absolutely free?

The products don't come with any management approval process right,
that would have to be built around the product right?

So what you are saying is, most new shops looking to plug in free
change control would likely choose one of those new and improved
version, there would be no good reason to go to CVS?

Thanks again.

On 2004-04-16, jason@cyberpine.com <jason@cyberpine.com> wrote:
> Andre Majorel <amajorel@teezer.fr> wrote in message >
>> Subversion is meant as the successor to CVS, without the
>> limitations of the original (the lack of support for directories
>> and symlinks, renaming, atomic commits, changesets, undo, etc.)
>>
>> GNU arch is more powerful than Subversion, but also more
>> complex.
>>
>> CVS became very popular because it was more or less the only
>> libre SCM allowing concurrent access. Now that better
>> alternatives are available, I expect that most CVS users will
>> gradually move to either Subversion or GNU arch. My employer (a
>> Unix ISV) plans to do so in the short term.
>
> Thanks for your response, very helpful right now.
>
> Some newbie questions
>
> Are those limitation still true for CVS?

Yes, and they always will be. As CVS is basically a wrapper
around RCS, it inherits all the limitations of RCS. Most of the
limitations I mentioned are structural and cannot be fixed while
maintaining compatibility. There have been several protracted
threads about this on the CVS mailing list.

> Lets say I wanted to encapsulate the promotion process in a script and
> test that sources for certain code standards and reject the promotion
> if that naming convention or actual code standard fails? Possible with
> all three products? Possible in likely the same ways?
>
> With such a large install base for CVS, would it not be safer (from a
> support point of view) to stay with the croud for now. Reason I ask is
> that management is hesitant to go freeware unless we can prove that
> the list servers and online doc are as good as or better than phone
> support you get with a commercial product.

Both projects have fewer users, but they have mailing lists,
just like CVS does.

I know where you're coming from. Sure, CVS is easier to sell to
the management because it's proven, and it has a lot of users.
But I don't think it's wise to bet on the current situation,
because the situation is not static. My bold prediction is that
a couple years from now, nobody will use CVS anymore, at least
not for new projects, and you'll be stuck with an obsolete VC
system.

I think the only relevant question is : are GNU arch and
Subversion usable *now* ? Read what their users say on their
mailing list archives and make your own opinion.

> All three are absolutely free?

Yes.

> The products don't come with any management approval process right,
> that would have to be built around the product right?

Sorry, I'm out of my depth here.

> So what you are saying is, most new shops looking to plug in free
> change control would likely choose one of those new and improved
> version, there would be no good reason to go to CVS?

Yes, for a new project in a shop that has no past CVS
experience, I don't see any reason to go for CVS instead of
Subversion (unless Subversion is still too buggy, but that's not
what I've heard).

GNU arch is a different animal. It's more like bitkeeper than
CVS in scope.

--
André Majorel <URL:http://www.teaser.fr/~amajorel/>
"Finally I am becoming stupider no more." -- Paul Erdös' epitaph