version control using subversion

[stroller]

hi all, i'm working on a new skill based app to integrate subversion
with virtuoso/composer. i noticed alot of the previous solutions like
perforce used the GDM (generic data management) interface to ease the
integration but my salesguy friend at cadence can't find gdm in the
book anymore. should i be trying to use this?

also, i'd be interested in hearing from anyone that has tried this
integration themselves. might be interesting to compare notes.

thx!

 

[John Gianni]

stroller wrote:

hi all, i'm working on a new skill based app to integrate subversion
with virtuoso/composer. i noticed alot of the previous solutions like
perforce used the GDM (generic data management) ...

<soapbox>
Data management isn't a tool; it's a religion.
</soapbox>

Having said that, the GDM interface, as far as I remember, isn't
something you want to work with if you're not a licensed business
partner with Cadence.

What Cadence recommends, AFAIR, at least on the Virtuoso (aka DFII)
tool suite, is you either employ the DM tools provided by Cadence
(depending on the version of your software) or by Matrix One (formerly
Syncronicity).

I'll repeat though, since everyone does DM differently ... DM is more
about the religion & enforcement of use models than it is about tools.
My recommendation is to choose an existing DM supplier recommended by
your Cadence sales and support team and then choose one of the use
models that DM supplier recommends and supports.

John Gianni
-- Nothing stated by me on the USENET is previously vetted by my
employer.

 

[fogh]

John Gianni wrote:

stroller wrote:

hi all, i'm working on a new skill based app to integrate subversion
with virtuoso/composer. i noticed alot of the previous solutions like
perforce used the GDM (generic data management) ...


soapbox
Data management isn't a tool; it's a religion.
/soapbox

At last a hope that I am not alone. It s hard being an atheist sometimes.

 

[jbrand]

I looked into this a few years ago. There were three options for data
management that I found.

1) Versionsync - already integrated into cadence
2) ClioSoft
3) Synchronicity

The bottom rung was Versionsync which has basic funtionality and the
top was Synchronicity. Now mind you that this is my opinion.

We ended up writing skill around versionsync. It ended up being a
hassle and we eventually killed the idea.

Joe

 

[Svenn Are Bjerkem]

fogh wrote:

At last a hope that I am not alone.  It s hard being an atheist sometimes.

Not quite sure if not atheism is a kind of religion as it is as filled with
believers as real religions.

Biggest problem I see with revision management and actually cadence as such
is that real concurrent work is not really possible with most of the
management systems as they do hard locking on the files. Propagation of
changes throughout a designer community is very often a hassle and you
cannot just go on and change data because you don't know the implications
across the entire tree. Cadence as a tool relies on other tools to fix
something that has been not handled very good within the framework.

The problem could be solved by using text based storage formats (XML) which
can be controlled by tools known to work in text based worlds with
utilities known by many people and free as in speach. But instead of
keeping things simple and stupid, there are dinosaur solutions with vobs
and vaults and such scaring the shit out of anybody not having 100+ hours
of training on the beast.

My religion is KIFF: Keep it free, finehead
--
Svenn

 

[fogh]

Svenn Are Bjerkem wrote:

fogh wrote:


At last a hope that I am not alone. It s hard being an atheist sometimes.

Not quite sure if not atheism is a kind of religion as it is as filled with
believers as real religions.

Biggest problem I see with revision management and actually cadence as such
is that real concurrent work is not really possible with most of the
management systems as they do hard locking on the files. Propagation of
changes throughout a designer community is very often a hassle and you
cannot just go on and change data because you don't know the implications
across the entire tree. Cadence as a tool relies on other tools to fix
something that has been not handled very good within the framework.

The problem could be solved by using text based storage formats (XML) which
can be controlled by tools known to work in text based worlds with
utilities known by many people and free as in speach. But instead of
keeping things simple and stupid, there are dinosaur solutions with vobs
and vaults and such scaring the shit out of anybody not having 100+ hours
of training on the beast.

My religion is KIFF: Keep it free, finehead

My religion is rather BAFF: be alone and face the fear. Last I heard
it is has no clergy.

So far everyone agree that synchronicity is top of the pops. As you
pointed out, it lacks 2 elementary things:
- broadcast to users of a module that it's current version has changed
- allow for diff and merge.

The latter does not really need XML , a schematic-versus-schematic
would be already a big step.
The former is plain lack feature and integration in dfII.

I d say a well agreed set of cds.lib's and scripts to enforce they are
sane more useful than any RCS. Some applets that help designers get in
touch (email owner of this module/lock, show phone number, ...) are
helping more than X choices on how to check out.

 

[Svenn Are Bjerkem]

fogh wrote:

I d say a well agreed set of cds.lib's and scripts to enforce they are
sane more useful than any RCS. Some applets that help designers get in
touch (email owner of this module/lock, show phone number, ...) are
helping more than X choices on how to check out.

Yeah, let's implement an interface to an instant messaging system. (In my
environment it must be something windows based because management does not
understand unix). Then people can decide themselves if they want to listen
or not. Won't litter up the mailbox either.

--
Svenn

 

[fogh]

Svenn Are Bjerkem wrote:

fogh wrote:


I d say a well agreed set of cds.lib's and scripts to enforce they are
sane more useful than any RCS. Some applets that help designers get in
touch (email owner of this module/lock, show phone number, ...) are
helping more than X choices on how to check out.


Yeah, let's implement an interface to an instant messaging system. (In my
environment it must be something windows based because management does not
understand unix). Then people can decide themselves if they want to listen
or not. Won't litter up the mailbox either.

If you want to cross boundary between unix and windows, you may find
complications and need system administrators help anyway...
There is in Samba a way to send winpopup notifications like those
defined in windows for workgroups. But that already requires that:
- the unix machine with samba is authenticated in the MS "domain
controller", and that the samba setup is complete enough for that
- the windows machines with the users interested can be enumerated. The
first step of course is that you can map the unix user to MSwindows
user. There you can make use of NIS or LDAP or SMB/NMB.

Maybe you can arrange something with GAIM or mozilla or some instant
messaging system that is easier on the authentication side. I have no
experience in scripting that kind of application, tho.

 

[Svenn Are Bjerkem]

fogh wrote:

Maybe you can arrange something with GAIM or mozilla or some instant
messaging system that is easier on the authentication side. I have no
experience in scripting that kind of application, tho.

We just started using VoIP and there is a connection from the PBX to the
messenger that comes with windows. Some quite nice features for those who
bother to use it. I could not impress any colleacues with dinner
invitations on IM, so I would expect some resistance vs. anything new. On
the other side, I had an "interesting" discussion on revision control and
so called public and private workspaces the other day. Looks like most
Hardware designers enter the trap that cvs manual warns about very early:
Revision management does not replace the need for team communication.

I tried to get onto the internal mailinglist for designsync changes of the
project I work on, but got the message that I would probably not like the
high traffic that this would cause. I find it fine when design support
knows up ahead what I want.

I have to sleep on the idea for some time. I have too many to handle
already.
--
Svenn