You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Mike Mason <mg...@thoughtworks.net> on 2004/01/07 07:43:36 UTC
Subversion vs. CVS (was Re: Congratulations!)
kfogel@collab.net wrote:
>
>>Since 1.0 is now just around the corner, I wonder if someone on the
>>SVN team would be interested in writing a brief article on the "event"
>>to post on the cmcrossroads.com announcements list? If someone would
>>even be willing to write a limited series of small articles on SVN
>>concepts and operation and how it improves CVS, I can ask the editor
>>if room could be made to accommodate such a "series" because I feel
>>this is a huge event that an extremely large community (all CVS users
>>at a bare minimum) will be interested in.
>>
>>
>
>That would be great... I've cross-posted this to the dev list, since
>there are potential authors there too.
>
>
I'd be happy to help write (or contribute to) anything that helps
evangelise Subversion. I've been mentioning Subversion on my blog every
now and then, even incurring the displeasure of a certain source control
vendor, so putting together something like this sounds fun.
Can we brainstorm a list of things that we'd like to mention when
talking about Subversion for CVS users?
- designed from scratch, by people who know what they're talking about
(anyone want to pipe up with credentials? Several people here "wrote the
book" on CVS, I know this as it's sitting on my desk...)
- high-performance database backend, not a collection of individual RCS
files
- branching and tagging as cheap operations
- atomic commit (include discussion about *why* this is useful!)
- optionally leverages Apache for networking (why this is good, plus
talk about svn/svn+ssh)
- excellent internals, leading to good APIs for add-on tools (give
examples of tools that exist, and maybe some that don't exist for CVS
which would be feasible for Subversion)
What have I missed? What CVS misfeatures does Subversion fix that really
help sell it?
A lot of times I hear, "Subversion, 3 years in development, and it's
nothing new" -- I happen to disagree, I think Perforce is the only thing
that comes close (that I've used) to having this set of useful features.
No, it doesn't do distributed this that and the other -- but do you
*need* that? I need decent, simple, everyday operation from source
control, which lets me control what I do, not trip up other developers,
and helps me figure out what's happened to the code through decent
history tracking. This probably isn't something to mention directly when
talking about how great we think Subversion is, I just wanted to get it
off my chest!
Is there also a need for a comparison between Subversion and various
other source control systems? I've read a few, including some which had
done plenty of research and included feature-list tickboxes, but that
read a little bit like the author hadn't used much of the systems they
were comparing to. For example, I'd feel qualified to compare Subversion
to CVS, Perforce, and StarTeam (and, to be honest, I still think
Perforce is better, although not in every situation). The Perforce
website includes a page with interesting performance benchmarks on it,
and I think it would be good for Subversion to have that kind of
reference too.
Cheers,
Mike.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by Emmeran Seehuber <ro...@rototor.de>.
On Wednesday 07 January 2004 07:43, Mike Mason wrote:
> Can we brainstorm a list of things that we'd like to mention when
> talking about Subversion for CVS users?
- You get an implicit "label" with each commit (i.e. the revision number). If
you commit all your changes which e.g. implement a feature or fix a bug at
once, you can easily track down all changes in the different files needed for
that feature/bug. It's a little bit like a "ChangeSet". This hardly possible
with CVS or VSS.
cu,
Emmy
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by "Jason E. Stewart" <ja...@openinformatics.com>.
kfogel@collab.net writes:
> jason@openinformatics.com (Jason E. Stewart) writes:
> > One of the big issues is how efficient SVN handles networking. I
> > travel all over the world and often have very tiny pipes with huge
> > latencies - CVS is impossible even for trivial tasks, whereas SVN is
> > effortless even for large commits/changes.
>
> That's really nice to hear.
>
> At work, I have a high-bandwidth, generally low-latency connection, so
> never see this advantnge. At home, DSL, not as fast as work but still
> pretty good. It's nice to know that the theory of SVN winning over
> tiny pipes turns out to be true in practice.
Because of my travels I've forced a few projects to move from CVS to
SVN. I'm hoping that SourceForge will soon adopt (or at least enable)
SVN for its projects as well.
SVN has made travelling the world as an Open Source developer possible
- without it, I'd have had to find a new way of making a living.
Thanks for an awesome piece of work,
jas.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by "Jason E. Stewart" <ja...@openinformatics.com>.
kfogel@collab.net writes:
> jason@openinformatics.com (Jason E. Stewart) writes:
> > One of the big issues is how efficient SVN handles networking. I
> > travel all over the world and often have very tiny pipes with huge
> > latencies - CVS is impossible even for trivial tasks, whereas SVN is
> > effortless even for large commits/changes.
>
> That's really nice to hear.
>
> At work, I have a high-bandwidth, generally low-latency connection, so
> never see this advantnge. At home, DSL, not as fast as work but still
> pretty good. It's nice to know that the theory of SVN winning over
> tiny pipes turns out to be true in practice.
Because of my travels I've forced a few projects to move from CVS to
SVN. I'm hoping that SourceForge will soon adopt (or at least enable)
SVN for its projects as well.
SVN has made travelling the world as an Open Source developer possible
- without it, I'd have had to find a new way of making a living.
Thanks for an awesome piece of work,
jas.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by kf...@collab.net.
jason@openinformatics.com (Jason E. Stewart) writes:
> > What have I missed? What CVS misfeatures does Subversion fix that really
> > help sell it?
>
> One of the big issues is how efficient SVN handles networking. I
> travel all over the world and often have very tiny pipes with huge
> latencies - CVS is impossible even for trivial tasks, whereas SVN is
> effortless even for large commits/changes.
That's really nice to hear.
At work, I have a high-bandwidth, generally low-latency connection, so
never see this advantnge. At home, DSL, not as fast as work but still
pretty good. It's nice to know that the theory of SVN winning over
tiny pipes turns out to be true in practice.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by kf...@collab.net.
jason@openinformatics.com (Jason E. Stewart) writes:
> > What have I missed? What CVS misfeatures does Subversion fix that really
> > help sell it?
>
> One of the big issues is how efficient SVN handles networking. I
> travel all over the world and often have very tiny pipes with huge
> latencies - CVS is impossible even for trivial tasks, whereas SVN is
> effortless even for large commits/changes.
That's really nice to hear.
At work, I have a high-bandwidth, generally low-latency connection, so
never see this advantnge. At home, DSL, not as fast as work but still
pretty good. It's nice to know that the theory of SVN winning over
tiny pipes turns out to be true in practice.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by "Jason E. Stewart" <ja...@openinformatics.com>.
Mike Mason <mg...@thoughtworks.net> writes:
> What have I missed? What CVS misfeatures does Subversion fix that really
> help sell it?
One of the big issues is how efficient SVN handles networking. I
travel all over the world and often have very tiny pipes with huge
latencies - CVS is impossible even for trivial tasks, whereas SVN is
effortless even for large commits/changes.
Cheers,
jas.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by Mike Mason <mg...@thoughtworks.net>.
John Peacock wrote:
> I don't know when you last tried it, but CVSNT is very capable, under
> active development, and supports features that 'base' CVS doesn't
> (like mergepoint processing, which svn doesn't yet handle). CVSNT
> supports multiple authentication methods (SSL, SSPI, etc) and for at
> least a year has had a dedicated lock server (no more lockfiles). The
> server process runs very well under NT4-W2K and the server code
> actually builds and runs under Linux as well (with limited SSPI support).
I tried CVSNT four months ago before deciding to go with Subversion.
Basically, I was trying to do a fairly hefty initial import, and CVSNT
couldn't handle it, giving me hangs during import. I didn't try too hard
to get it to work, though. Subversion worked great first time so I used
that.
Cheers,
Mike.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
RE: Subversion vs. CVS (was Re: Congratulations!)
Posted by Richard Birkby <sv...@thundermain.com>.
John Peacock wrote:
>
> J Wynia wrote:
>
> > To me, one of the biggest benefits (and one that doesn't tend to get
> > mentioned) is that it's cross platform from the beginning.
> While there are
> > CVS clients, running a CVS server on Windows is nearly
> impossible and even
> > attempting it brings comments like "Don't even try" from the "CVS
> > experts".
>
> I don't know when you last tried it, but CVSNT is very capable,
I've tried it and failed. Subversion is childs play to set up in comparison.
Richard
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by John Peacock <jp...@rowman.com>.
J Wynia wrote:
> To me, one of the biggest benefits (and one that doesn't tend to get
> mentioned) is that it's cross platform from the beginning. While there are
> CVS clients, running a CVS server on Windows is nearly impossible and even
> attempting it brings comments like "Don't even try" from the "CVS
> experts".
I don't know when you last tried it, but CVSNT is very capable, under active
development, and supports features that 'base' CVS doesn't (like mergepoint
processing, which svn doesn't yet handle). CVSNT supports multiple
authentication methods (SSL, SSPI, etc) and for at least a year has had a
dedicated lock server (no more lockfiles). The server process runs very well
under NT4-W2K and the server code actually builds and runs under Linux as well
(with limited SSPI support).
If I didn't like subversion's re-engineered model, I wouldn't be switching away
from CVSNT in the first place...
http://www.cvsnt.org
HTH
John
--
John Peacock
Director of Information Research and Technology
Rowman & Littlefield Publishing Group
4501 Forbes Boulevard
Suite H
Lanham, MD 20706
301-459-3366 x.5010
fax 301-429-5748
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by J Wynia <jw...@pragmapool.com>.
To me, one of the biggest benefits (and one that doesn't tend to get
mentioned) is that it's cross platform from the beginning. While there are
CVS clients, running a CVS server on Windows is nearly impossible and even
attempting it brings comments like "Don't even try" from the "CVS
experts". The simple reality is that there's a LOT of development done in
a Windows environment and having to set up a Linux machine just for CVS
use has historically been out of the question in many places I've worked.
However, co-opting a Windows workstation is often a possibility. And, with
the file:// access methods, SVN can be used directly on your own
workstation for single developer work.
Ben Collins-Sussman said:
> On Wed, 2004-01-07 at 01:43, Mike Mason wrote:
>
>> A lot of times I hear, "Subversion, 3 years in development, and it's
>> nothing new" --
>
> This is pretty common FUD, I hear it a lot too.
>
> But I don't think it's even worth acknowledging or debating that
> criticism. The people who say such things are missing the whole point
> of Subversion; they disagree with what I call the "fundamental axiom"
> of the project:
>
> CVS is a good model for version control,
> it just wasn't implemented well.
>
> Some people say Subversion is just "polishing a turd", but my take is
> that we're polishing a "diamond in the rough". We take the CVS model
> and add directory versioning, atomic commits, database backend,
> properties, efficient binary handling, flexible network abilities, and a
> solid C API. It's what CVS should have been in the first place.
>
> But if someone doesn't like the CVS model to begin with, there's just no
> point in trying to sell them Subversion. Point them to arch, monotone,
> svk, aegis, or some other system with a different model.
>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
--
J Wynia
Pragmapool, Inc.
www.pragmapool.com
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by Ben Collins-Sussman <su...@collab.net>.
On Wed, 2004-01-07 at 01:43, Mike Mason wrote:
> A lot of times I hear, "Subversion, 3 years in development, and it's
> nothing new" --
This is pretty common FUD, I hear it a lot too.
But I don't think it's even worth acknowledging or debating that
criticism. The people who say such things are missing the whole point
of Subversion; they disagree with what I call the "fundamental axiom"
of the project:
CVS is a good model for version control,
it just wasn't implemented well.
Some people say Subversion is just "polishing a turd", but my take is
that we're polishing a "diamond in the rough". We take the CVS model
and add directory versioning, atomic commits, database backend,
properties, efficient binary handling, flexible network abilities, and a
solid C API. It's what CVS should have been in the first place.
But if someone doesn't like the CVS model to begin with, there's just no
point in trying to sell them Subversion. Point them to arch, monotone,
svk, aegis, or some other system with a different model.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: Subversion vs. CVS (was Re: Congratulations!)
Posted by Ben Collins-Sussman <su...@collab.net>.
On Wed, 2004-01-07 at 01:43, Mike Mason wrote:
> A lot of times I hear, "Subversion, 3 years in development, and it's
> nothing new" --
This is pretty common FUD, I hear it a lot too.
But I don't think it's even worth acknowledging or debating that
criticism. The people who say such things are missing the whole point
of Subversion; they disagree with what I call the "fundamental axiom"
of the project:
CVS is a good model for version control,
it just wasn't implemented well.
Some people say Subversion is just "polishing a turd", but my take is
that we're polishing a "diamond in the rough". We take the CVS model
and add directory versioning, atomic commits, database backend,
properties, efficient binary handling, flexible network abilities, and a
solid C API. It's what CVS should have been in the first place.
But if someone doesn't like the CVS model to begin with, there's just no
point in trying to sell them Subversion. Point them to arch, monotone,
svk, aegis, or some other system with a different model.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org