You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@subversion.apache.org by "Irvine, Chuck R [EQ]" <Ch...@Embarq.com> on 2007/04/10 15:46:09 UTC

Release Branches / 3-plus Development Streams

We have lately adopted SVN as our standard source code mgt tool. 

Previously, we used CVS and had a standard approach to branching in the
context of multiple concurrent release development streams. For two
concurrent streams, it was basically what is described in the SVN Book
as "Release Branches". Unfortunately, we are often forced to develop for
3 and 4 releases concurrently. (The SVN book does't comment specifically
on this situation.)

For this situation, our branching model using CVS was basically to
always have the latest release (R) on the trunk. Whenever we started on
release R+1, create a release branch for R and let R+1 development
proceed on the trunk. Graphically, it looks like this:


             /---------------------
            / R1
           /
          /            /---------------
         /            /  R2
        /            /
-------/------------/-------------------
R1        R2          R3

We would then merge from R1 to R2, from R2 to R3, etc. Eventually, R
would be deployed to production and R-1 would be retired. 

So, here's my question, so we don't go down the wrong track, can any of
the experienced SVN folks comment on the soundness of this approach
using SVN? I've been experimenting with it using SVN and it seems to
work nicely. Any input would be much appreciated. Thanks

Chuck Irvine
Embarq

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


RE: RE: Release Branches / 3-plus Development Streams

Posted by Ch...@qimonda.com.
I think this is why I like developing on a branch, rather
than on the trunk - it just seems more natural to me anyway.
We tag the trunk code for releases, and release the tag. 
Only bug fixes, and not features, are merged to the trunk.
As bug fixes are merged to trunk, they are subsequently
merged to all the branches. We run regression on trunk
code for sanitation.

>-----Original Message-----
>From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com] 
>Sent: Tuesday, April 10, 2007 12:20 PM
>To: Oefelein, Martina; users@subversion.tigris.org
>Subject: RE: Release Branches / 3-plus Development Streams
>
>Hi Martina,
>
>I doubt that merging from newer releases to older releases 
>would work for us, mainly for the reason that you allude to. 
>If we did that, we would be merging new features into old 
>releases (even into our current productiuon release), even 
>before the new features were robust and fully functional. The 
>other way around is almost always applicable. Any bugs found 
>in the production release, are merged down into streams still 
>in development. And, new and evolving functionality in 
>development stream R is merged appropriately into development 
>stream R+1. 
>
>Chuck
>
>> -----Original Message-----
>> From: Oefelein, Martina [mailto:Martina.Oefelein@dionex.com]
>> Sent: Tuesday, April 10, 2007 11:01 AM
>> To: Irvine, Chuck R [EQ]; users@subversion.tigris.org
>> Subject: RE: Release Branches / 3-plus Development Streams
>> 
>> 
>> 
>> Hi Chuck,
>> 
>> Like you, we have to maintain 3 existing releases concurrently with 
>> the trunk. We develop (usually) in the trunk and merge fixes (and 
>> sometimes features) to the release branches. If I understand you 
>> correctly, you are merging in the other direction, from the oldest 
>> release branch to the newer branches and finally to the trunk.
>> 
>> I think both ways will work technically, and the choice is probably 
>> more a question of development philosophy and procedures.
>> 
>> ciao
>> Martina
>> 
>> 
>> 
>> -----Original Message-----
>> From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com]
>> Sent: Tuesday, April 10, 2007 5:46 PM
>> To: users@subversion.tigris.org
>> Subject: Release Branches / 3-plus Development Streams
>> 
>> We have lately adopted SVN as our standard source code mgt tool. 
>> 
>> Previously, we used CVS and had a standard approach to branching in 
>> the context of multiple concurrent release development streams. For 
>> two concurrent streams, it was basically what is described 
>in the SVN 
>> Book as "Release Branches". Unfortunately, we are often forced to 
>> develop for
>> 3 and 4 releases concurrently. (The SVN book does't comment 
>> specifically on this situation.)
>> 
>> For this situation, our branching model using CVS was basically to 
>> always have the latest release (R) on the trunk.
>> Whenever we started on release R+1, create a release branch 
>for R and 
>> let R+1 development proceed on the trunk.
>> Graphically, it looks like this:
>> 
>> 
>>              /---------------------
>>             / R1
>>            /
>>           /            /---------------
>>          /            /  R2
>>         /            /
>> -------/------------/-------------------
>> R1        R2          R3
>> 
>> We would then merge from R1 to R2, from R2 to R3, etc. 
>> Eventually, R would be deployed to production and R-1 would be 
>> retired.
>> 
>> So, here's my question, so we don't go down the wrong track, can any 
>> of the experienced SVN folks comment on the soundness of 
>this approach 
>> using SVN? I've been experimenting with it using SVN and it seems to 
>> work nicely. Any input would be much appreciated. Thanks
>> 
>> Chuck Irvine
>> Embarq
>> 
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
>> For additional commands, e-mail: users-help@subversion.tigris.org
>> 
>> 
>> 
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
>For additional commands, e-mail: users-help@subversion.tigris.org
>

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


RE: Release Branches / 3-plus Development Streams

Posted by "Irvine, Chuck R [EQ]" <Ch...@Embarq.com>.
Hi Martina,

I doubt that merging from newer releases to older releases would work
for us, mainly for the reason that you allude to. If we did that, we
would be merging new features into old releases (even into our current
productiuon release), even before the new features were robust and fully
functional. The other way around is almost always applicable. Any bugs
found in the production release, are merged down into streams still in
development. And, new and evolving functionality in development stream R
is merged appropriately into development stream R+1. 

Chuck

> -----Original Message-----
> From: Oefelein, Martina [mailto:Martina.Oefelein@dionex.com] 
> Sent: Tuesday, April 10, 2007 11:01 AM
> To: Irvine, Chuck R [EQ]; users@subversion.tigris.org
> Subject: RE: Release Branches / 3-plus Development Streams
> 
> 
> 
> Hi Chuck,
> 
> Like you, we have to maintain 3 existing releases 
> concurrently with the trunk. We develop (usually) in the 
> trunk and merge fixes (and sometimes features) to the release 
> branches. If I understand you correctly, you are merging in 
> the other direction, from the oldest release branch to the 
> newer branches and finally to the trunk. 
> 
> I think both ways will work technically, and the choice is 
> probably more a question of development philosophy and procedures.
> 
> ciao 
> Martina 
> 
> 
> 
> -----Original Message-----
> From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com] 
> Sent: Tuesday, April 10, 2007 5:46 PM
> To: users@subversion.tigris.org
> Subject: Release Branches / 3-plus Development Streams
> 
> We have lately adopted SVN as our standard source code mgt tool. 
> 
> Previously, we used CVS and had a standard approach to 
> branching in the context of multiple concurrent release 
> development streams. For two concurrent streams, it was 
> basically what is described in the SVN Book as "Release 
> Branches". Unfortunately, we are often forced to develop for 
> 3 and 4 releases concurrently. (The SVN book does't comment 
> specifically on this situation.)
> 
> For this situation, our branching model using CVS was 
> basically to always have the latest release (R) on the trunk. 
> Whenever we started on release R+1, create a release branch 
> for R and let R+1 development proceed on the trunk. 
> Graphically, it looks like this:
> 
> 
>              /---------------------
>             / R1
>            /
>           /            /---------------
>          /            /  R2
>         /            /
> -------/------------/-------------------
> R1        R2          R3
> 
> We would then merge from R1 to R2, from R2 to R3, etc. 
> Eventually, R would be deployed to production and R-1 would 
> be retired. 
> 
> So, here's my question, so we don't go down the wrong track, 
> can any of the experienced SVN folks comment on the soundness 
> of this approach using SVN? I've been experimenting with it 
> using SVN and it seems to work nicely. Any input would be 
> much appreciated. Thanks
> 
> Chuck Irvine
> Embarq
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
> 
> 
> 

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


RE: Release Branches / 3-plus Development Streams

Posted by "Oefelein, Martina" <Ma...@dionex.com>.
Hi Chuck,

Like you, we have to maintain 3 existing releases concurrently with the trunk. We develop (usually) in the trunk and merge fixes (and sometimes features) to the release branches. If I understand you correctly, you are merging in the other direction, from the oldest release branch to the newer branches and finally to the trunk. 

I think both ways will work technically, and the choice is probably more a question of development philosophy and procedures.

ciao 
Martina 



-----Original Message-----
From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com] 
Sent: Tuesday, April 10, 2007 5:46 PM
To: users@subversion.tigris.org
Subject: Release Branches / 3-plus Development Streams

We have lately adopted SVN as our standard source code mgt tool. 

Previously, we used CVS and had a standard approach to branching in the
context of multiple concurrent release development streams. For two
concurrent streams, it was basically what is described in the SVN Book
as "Release Branches". Unfortunately, we are often forced to develop for
3 and 4 releases concurrently. (The SVN book does't comment specifically
on this situation.)

For this situation, our branching model using CVS was basically to
always have the latest release (R) on the trunk. Whenever we started on
release R+1, create a release branch for R and let R+1 development
proceed on the trunk. Graphically, it looks like this:


             /---------------------
            / R1
           /
          /            /---------------
         /            /  R2
        /            /
-------/------------/-------------------
R1        R2          R3

We would then merge from R1 to R2, from R2 to R3, etc. Eventually, R
would be deployed to production and R-1 would be retired. 

So, here's my question, so we don't go down the wrong track, can any of
the experienced SVN folks comment on the soundness of this approach
using SVN? I've been experimenting with it using SVN and it seems to
work nicely. Any input would be much appreciated. Thanks

Chuck Irvine
Embarq

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


RE: RE: RE: RE: Release Branches / 3-plus Development Streams

Posted by Ch...@qimonda.com.
Yes, this is currently what we're doing NOW, in my current
company with SVN. We have development branches, and merge
from R1-to-trunk, then trunk-to-R2, and ..to-R3 if it exists.
Note in addition to branches, we create tags which ease or
merging. Read up on tags if you have not yet.

>-----Original Message-----
>From: Daniell, Casey B [mailto:Casey_Daniell@reyrey.com] 
>Sent: Monday, April 23, 2007 10:51 AM
>To: Irvine, Chuck R [EQ]; users@subversion.tigris.org
>Subject: RE: RE: RE: Release Branches / 3-plus Development Streams
>
>The trunk is really being used as a point to merge between releases. 
>
>You are correct in assuming that by "pristine" I do not mean 
>totally unchanged, but rather, only the Config Managers can 
>make changes on this branch -- all developers must commit on a 
>branch. By doing this we keep it easy to determine on the 
>initial merges to the TRUNK, what changes should be kept 
>should a conflict arise.  
>
>I think that answers your questions...
>
>Casey
>
>-----Original Message-----
>From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com]
>Sent: Thursday, April 19, 2007 5:27 PM
>To: Daniell, Casey B; users@subversion.tigris.org
>Subject: RE: RE: RE: Release Branches / 3-plus Development Streams
>
>I don't understand a couple of things about branch/merge 
>scheme. If all releases are developmented on branches, what is 
>the trunk really being used for? Secondly, what do you mean 
>that you keep the trunk pristine?
>By pristine I might think you mean "unchanged" but that can't 
>be right since you are merging changes to the trunk
>
>Chuck
>
>> -----Original Message-----
>> From: Daniell, Casey B [mailto:Casey_Daniell@reyrey.com]
>> Sent: Thursday, April 19, 2007 2:30 PM
>> To: users@subversion.tigris.org
>> Subject: RE: RE: Release Branches / 3-plus Development Streams
>> 
>> 
>> I am currently doing almost what your propose, instead we keep the 
>> TRUNK pristine and have all development occur on the R 
>branches. Basic
>
>> model
>> is:
>> 
>> Development occurs on the R1 branch and when done, or at a point we 
>> need to promote the changes to other release(s), I merge the deltas 
>> from the start of the branch to the end to the TRUNK. During this 
>> merge since all changes by DEV has occurred on the branch I know to 
>> take whatever changes occurred on the branch.
>> 
>> Then, from the TRUNK I merge the one new revision created 
>from the R1 
>> to TRUNK merge to the R2 branch. If there was an
>> R3 branch created at the same times, changes would go out to them as 
>> well. If you do this outer merge you will have to be careful when 
>> merge the R2 and R3 branches to the TRUNK. See the SVN docs 
>on how to 
>> do this merge, but you will be picking the rev that the 
>TRUNK and R2, 
>> or R3 branch sync'd and the last revision on the branch, 
>then merge to
>
>> the target location. (AKA probably the TRUNK again)...
>> 
>> Casey
>> 
>> -----Original Message-----
>> From: Chris.Fouts@qimonda.com [mailto:Chris.Fouts@qimonda.com]
>> Sent: Thursday, April 19, 2007 10:18 AM
>> To: users@subversion.tigris.org
>> Subject: RE: Release Branches / 3-plus Development Streams
>> 
>> I worked for a company that also did this, and with CVS.
>> In that case,
>> - R+1 was never merged to R
>> - any fix for R+1 also went to R, if applicable
>> - any fix for R also went to R+1, if applicable
>> 
>> Since SVN and CVS are basically similar revision control systems as 
>> far as ideology is concerned, I don't see how SVN can change the 
>> approach you have now.
>> 
>> >-----Original Message-----
>> >From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com]
>> >Sent: Thursday, April 19, 2007 11:05 AM
>> >To: users@subversion.tigris.org
>> >Subject: Release Branches / 3-plus Development Streams
>> >
>> >This is a repeat of a previous mailing list posting. Pardon
>> the repeat,
>> 
>> >but no one responded and, since it is a pretty question for us, I 
>> >thouht that I would try again. Thanks.
>> >
>> >.....
>> >
>> >We have lately adopted SVN as our standard source code mgt tool.
>> >
>> >Previously, we used CVS and had a standard approach to
>> branching in the
>> 
>> >context of multiple concurrent release development streams. For two 
>> >concurrent streams, it was basically what is described in
>> the SVN Book
>> >as "Release Branches". Unfortunately, we are often forced 
>to develop 
>> >for
>> >3 and 4 releases concurrently. (The SVN book does't comment 
>> >specifically on this situation.)
>> >
>> >For this situation, our branching model using CVS was basically to 
>> >always have the latest release (R) on the trunk.
>> >Whenever we started on release R+1, create a release branch
>> for R and
>> >let R+1 development proceed on the trunk.
>> >Graphically, it looks like this:
>> >
>> >
>> >             /---------------------
>> >            / R1
>> >           /
>> >          /            /---------------
>> >         /            /  R2
>> >        /            /
>> >-------/------------/-------------------
>> >R1        R2          R3
>> >
>> >We would then merge from R1 to R2, from R2 to R3, etc.
>> >Eventually, R would be deployed to production and R-1 would be 
>> >retired.
>> >
>> >So, here's my question, so we don't go down the wrong 
>track, can any 
>> >of the experienced SVN folks comment on the soundness of this 
>> >approach using SVN? I've been experimenting with it using 
>SVN and it 
>> >seems to work nicely. Any input would be much appreciated. Thanks
>> >
>> >Chuck Irvine
>> >Embarq
>> >
>> >
>> 
>>---------------------------------------------------------------------
>> >To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
>> >For additional commands, e-mail: users-help@subversion.tigris.org
>> >
>> 
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
>> For additional commands, e-mail: users-help@subversion.tigris.org
>> 
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
>> For additional commands, e-mail: users-help@subversion.tigris.org
>> 
>> 
>> 
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
>For additional commands, e-mail: users-help@subversion.tigris.org
>

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


RE: RE: RE: Release Branches / 3-plus Development Streams

Posted by "Daniell, Casey B" <Ca...@reyrey.com>.
The trunk is really being used as a point to merge between releases. 

You are correct in assuming that by "pristine" I do not mean totally
unchanged, but rather, only the Config Managers can make changes on this
branch -- all developers must commit on a branch. By doing this we keep
it easy to determine on the initial merges to the TRUNK, what changes
should be kept should a conflict arise.  

I think that answers your questions...

Casey

-----Original Message-----
From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com] 
Sent: Thursday, April 19, 2007 5:27 PM
To: Daniell, Casey B; users@subversion.tigris.org
Subject: RE: RE: RE: Release Branches / 3-plus Development Streams

I don't understand a couple of things about branch/merge scheme. If all
releases are developmented on branches, what is the trunk really being
used for? Secondly, what do you mean that you keep the trunk pristine?
By pristine I might think you mean "unchanged" but that can't be right
since you are merging changes to the trunk

Chuck

> -----Original Message-----
> From: Daniell, Casey B [mailto:Casey_Daniell@reyrey.com]
> Sent: Thursday, April 19, 2007 2:30 PM
> To: users@subversion.tigris.org
> Subject: RE: RE: Release Branches / 3-plus Development Streams
> 
> 
> I am currently doing almost what your propose, instead we keep the 
> TRUNK pristine and have all development occur on the R branches. Basic

> model
> is:
> 
> Development occurs on the R1 branch and when done, or at a point we 
> need to promote the changes to other release(s), I merge the deltas 
> from the start of the branch to the end to the TRUNK. During this 
> merge since all changes by DEV has occurred on the branch I know to 
> take whatever changes occurred on the branch.
> 
> Then, from the TRUNK I merge the one new revision created from the R1 
> to TRUNK merge to the R2 branch. If there was an
> R3 branch created at the same times, changes would go out to them as 
> well. If you do this outer merge you will have to be careful when 
> merge the R2 and R3 branches to the TRUNK. See the SVN docs on how to 
> do this merge, but you will be picking the rev that the TRUNK and R2, 
> or R3 branch sync'd and the last revision on the branch, then merge to

> the target location. (AKA probably the TRUNK again)...
> 
> Casey
> 
> -----Original Message-----
> From: Chris.Fouts@qimonda.com [mailto:Chris.Fouts@qimonda.com]
> Sent: Thursday, April 19, 2007 10:18 AM
> To: users@subversion.tigris.org
> Subject: RE: Release Branches / 3-plus Development Streams
> 
> I worked for a company that also did this, and with CVS.
> In that case,
> - R+1 was never merged to R
> - any fix for R+1 also went to R, if applicable
> - any fix for R also went to R+1, if applicable
> 
> Since SVN and CVS are basically similar revision control systems as 
> far as ideology is concerned, I don't see how SVN can change the 
> approach you have now.
> 
> >-----Original Message-----
> >From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com]
> >Sent: Thursday, April 19, 2007 11:05 AM
> >To: users@subversion.tigris.org
> >Subject: Release Branches / 3-plus Development Streams
> >
> >This is a repeat of a previous mailing list posting. Pardon
> the repeat,
> 
> >but no one responded and, since it is a pretty question for us, I 
> >thouht that I would try again. Thanks.
> >
> >.....
> >
> >We have lately adopted SVN as our standard source code mgt tool.
> >
> >Previously, we used CVS and had a standard approach to
> branching in the
> 
> >context of multiple concurrent release development streams. For two 
> >concurrent streams, it was basically what is described in
> the SVN Book
> >as "Release Branches". Unfortunately, we are often forced to develop 
> >for
> >3 and 4 releases concurrently. (The SVN book does't comment 
> >specifically on this situation.)
> >
> >For this situation, our branching model using CVS was basically to 
> >always have the latest release (R) on the trunk.
> >Whenever we started on release R+1, create a release branch
> for R and
> >let R+1 development proceed on the trunk.
> >Graphically, it looks like this:
> >
> >
> >             /---------------------
> >            / R1
> >           /
> >          /            /---------------
> >         /            /  R2
> >        /            /
> >-------/------------/-------------------
> >R1        R2          R3
> >
> >We would then merge from R1 to R2, from R2 to R3, etc.
> >Eventually, R would be deployed to production and R-1 would be 
> >retired.
> >
> >So, here's my question, so we don't go down the wrong track, can any 
> >of the experienced SVN folks comment on the soundness of this 
> >approach using SVN? I've been experimenting with it using SVN and it 
> >seems to work nicely. Any input would be much appreciated. Thanks
> >
> >Chuck Irvine
> >Embarq
> >
> >
> >---------------------------------------------------------------------
> >To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> >For additional commands, e-mail: users-help@subversion.tigris.org
> >
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
> 
> 
> 

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


Re: Release Branches / 3-plus Development Streams

Posted by Les Mikesell <le...@gmail.com>.
Appleton, Will wrote:
> 
>> -----Original Message-----
>> From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com]
>> Sent: Thursday, April 19, 2007 5:27 PM
>> To: Daniell, Casey B; users@subversion.tigris.org
>> Subject: RE: RE: RE: Release Branches / 3-plus Development Streams
>>
>> I don't understand a couple of things about branch/merge scheme. If
> all
>> releases are developmented on branches, what is the trunk really being
>> used for? Secondly, what do you mean that you keep the trunk pristine?
>> By pristine I might think you mean "unchanged" but that can't be right
>> since you are merging changes to the trunk
>>
>> Chuck
> 
> Here's a nice article (treatise really) that has lots of branching
> strategies.  Nice clear explanations of various branching patterns and
> why/when to use them.  
> 
> http://www.cmcrossroads.com/bradapp/acme/branching/streamed-lines.html
> 
> Not really an answer to your question exactly, but it's nice background
> on branching/merging in general.

I think the first thing you need to decide is whether you care if the 
trunk/HEAD is always a released revision or not.  For some places, 
especially ones that permit public or uncontrolled access, this may be 
important.  If it isn't, it can make it easier to track history if you 
always develop on the trunk and only branch for releases that need their 
own separate maintenance and for changes that might be disruptive if not 
isolated from other concurrent work on the trunk.  You might not even 
need to create the release branches until the first support changes are 
needed.  Just copy any revision you want to be able to reproduce to a 
tag, then when you need to make a change that doesn't affect anything 
else, copy the tag to a branch to do the work.  I'm not sure you can 
generalize about what happens after this - sometimes the fix you need 
will already exist on the trunk and you can merge it, sometimes it will 
be something you don't need anywhere else.

-- 
   Les Mikesell
    lesmikesell@gmail.com


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

RE: Release Branches / 3-plus Development Streams

Posted by "Appleton, Will" <Wi...@sedgwickcms.com>.

> -----Original Message-----
> From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com]
> Sent: Thursday, April 19, 2007 5:27 PM
> To: Daniell, Casey B; users@subversion.tigris.org
> Subject: RE: RE: RE: Release Branches / 3-plus Development Streams
> 
> I don't understand a couple of things about branch/merge scheme. If
all
> releases are developmented on branches, what is the trunk really being
> used for? Secondly, what do you mean that you keep the trunk pristine?
> By pristine I might think you mean "unchanged" but that can't be right
> since you are merging changes to the trunk
> 
> Chuck

Here's a nice article (treatise really) that has lots of branching
strategies.  Nice clear explanations of various branching patterns and
why/when to use them.  

http://www.cmcrossroads.com/bradapp/acme/branching/streamed-lines.html

Not really an answer to your question exactly, but it's nice background
on branching/merging in general.

It sounds to me like y'all are discussing a variation of what is called
"Codeline Per Release" in the Streamed Lines article.

http://www.cmcrossroads.com/bradapp/acme/branching/branch-creation.html#
CodelinePerRelease


I hope this is an appropriate post - I'm new to this list. :)

Will Appleton
 



The information transmitted is intended only for the person or entity to which it is addressed and may contain confidential and/or privileged material. Any review, retransmission, dissemination or other use of, or taking of any action in reliance upon this information by persons or entities other than the intended recipient is prohibited. If you received this in error, please contact the sender and delete the material from any computer.

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


RE: RE: RE: Release Branches / 3-plus Development Streams

Posted by "Irvine, Chuck R [EQ]" <Ch...@Embarq.com>.
I don't understand a couple of things about branch/merge scheme. If all
releases are developmented on branches, what is the trunk really being
used for? Secondly, what do you mean that you keep the trunk pristine?
By pristine I might think you mean "unchanged" but that can't be right
since you are merging changes to the trunk

Chuck

> -----Original Message-----
> From: Daniell, Casey B [mailto:Casey_Daniell@reyrey.com] 
> Sent: Thursday, April 19, 2007 2:30 PM
> To: users@subversion.tigris.org
> Subject: RE: RE: Release Branches / 3-plus Development Streams
> 
> 
> I am currently doing almost what your propose, instead we 
> keep the TRUNK pristine and have all development occur on the 
> R branches. Basic model
> is:
> 
> Development occurs on the R1 branch and when done, or at a 
> point we need to promote the changes to other release(s), I 
> merge the deltas from the start of the branch to the end to 
> the TRUNK. During this merge since all changes by DEV has 
> occurred on the branch I know to take whatever changes 
> occurred on the branch. 
> 
> Then, from the TRUNK I merge the one new revision created 
> from the R1 to TRUNK merge to the R2 branch. If there was an 
> R3 branch created at the same times, changes would go out to 
> them as well. If you do this outer merge you will have to be 
> careful when merge the R2 and R3 branches to the TRUNK. See 
> the SVN docs on how to do this merge, but you will be picking 
> the rev that the TRUNK and R2, or R3 branch sync'd and the 
> last revision on the branch, then merge to the target 
> location. (AKA probably the TRUNK again)...
> 
> Casey
> 
> -----Original Message-----
> From: Chris.Fouts@qimonda.com [mailto:Chris.Fouts@qimonda.com] 
> Sent: Thursday, April 19, 2007 10:18 AM
> To: users@subversion.tigris.org
> Subject: RE: Release Branches / 3-plus Development Streams
> 
> I worked for a company that also did this, and with CVS.
> In that case,
> - R+1 was never merged to R
> - any fix for R+1 also went to R, if applicable
> - any fix for R also went to R+1, if applicable
> 
> Since SVN and CVS are basically similar revision control 
> systems as far as ideology is concerned, I don't see how SVN 
> can change the approach you have now.
> 
> >-----Original Message-----
> >From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com]
> >Sent: Thursday, April 19, 2007 11:05 AM
> >To: users@subversion.tigris.org
> >Subject: Release Branches / 3-plus Development Streams
> >
> >This is a repeat of a previous mailing list posting. Pardon 
> the repeat,
> 
> >but no one responded and, since it is a pretty question for us, I
> >thouht that I would try again. Thanks.
> >
> >.....
> >
> >We have lately adopted SVN as our standard source code mgt tool.
> >
> >Previously, we used CVS and had a standard approach to 
> branching in the
> 
> >context of multiple concurrent release development streams. For two
> >concurrent streams, it was basically what is described in 
> the SVN Book 
> >as "Release Branches". Unfortunately, we are often forced to develop 
> >for
> >3 and 4 releases concurrently. (The SVN book does't comment 
> >specifically on this situation.)
> >
> >For this situation, our branching model using CVS was basically to
> >always have the latest release (R) on the trunk.
> >Whenever we started on release R+1, create a release branch 
> for R and 
> >let R+1 development proceed on the trunk.
> >Graphically, it looks like this:
> >
> >
> >             /---------------------
> >            / R1
> >           /
> >          /            /---------------
> >         /            /  R2
> >        /            /
> >-------/------------/-------------------
> >R1        R2          R3
> >
> >We would then merge from R1 to R2, from R2 to R3, etc.
> >Eventually, R would be deployed to production and R-1 would be 
> >retired. 
> >
> >So, here's my question, so we don't go down the wrong track,
> >can any of the experienced SVN folks comment on the soundness 
> >of this approach using SVN? I've been experimenting with it 
> >using SVN and it seems to work nicely. Any input would be much 
> >appreciated. Thanks
> >
> >Chuck Irvine
> >Embarq
> >
> >
> >---------------------------------------------------------------------
> >To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> >For additional commands, e-mail: users-help@subversion.tigris.org
> >
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
> 
> 
> 

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


RE: RE: Release Branches / 3-plus Development Streams

Posted by "Daniell, Casey B" <Ca...@reyrey.com>.
I am currently doing almost what your propose, instead we keep the TRUNK
pristine and have all development occur on the R branches. Basic model
is:

Development occurs on the R1 branch and when done, or at a point we need
to promote the changes to other release(s), I merge the deltas from the
start of the branch to the end to the TRUNK. During this merge since all
changes by DEV has occurred on the branch I know to take whatever
changes occurred on the branch. 

Then, from the TRUNK I merge the one new revision created from the R1 to
TRUNK merge to the R2 branch. If there was an R3 branch created at the
same times, changes would go out to them as well. If you do this outer
merge you will have to be careful when merge the R2 and R3 branches to
the TRUNK. See the SVN docs on how to do this merge, but you will be
picking the rev that the TRUNK and R2, or R3 branch sync'd and the last
revision on the branch, then merge to the target location. (AKA probably
the TRUNK again)...

Casey

-----Original Message-----
From: Chris.Fouts@qimonda.com [mailto:Chris.Fouts@qimonda.com] 
Sent: Thursday, April 19, 2007 10:18 AM
To: users@subversion.tigris.org
Subject: RE: Release Branches / 3-plus Development Streams

I worked for a company that also did this, and with CVS.
In that case,
- R+1 was never merged to R
- any fix for R+1 also went to R, if applicable
- any fix for R also went to R+1, if applicable

Since SVN and CVS are basically similar revision control systems as far
as ideology is concerned, I don't see how SVN can change the approach
you have now.

>-----Original Message-----
>From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com]
>Sent: Thursday, April 19, 2007 11:05 AM
>To: users@subversion.tigris.org
>Subject: Release Branches / 3-plus Development Streams
>
>This is a repeat of a previous mailing list posting. Pardon the repeat,

>but no one responded and, since it is a pretty question for us, I 
>thouht that I would try again. Thanks.
>
>.....
>
>We have lately adopted SVN as our standard source code mgt tool. 
>
>Previously, we used CVS and had a standard approach to branching in the

>context of multiple concurrent release development streams. For two 
>concurrent streams, it was basically what is described in the SVN Book 
>as "Release Branches". Unfortunately, we are often forced to develop 
>for
>3 and 4 releases concurrently. (The SVN book does't comment 
>specifically on this situation.)
>
>For this situation, our branching model using CVS was basically to 
>always have the latest release (R) on the trunk.
>Whenever we started on release R+1, create a release branch for R and 
>let R+1 development proceed on the trunk.
>Graphically, it looks like this:
>
>
>             /---------------------
>            / R1
>           /
>          /            /---------------
>         /            /  R2
>        /            /
>-------/------------/-------------------
>R1        R2          R3
>
>We would then merge from R1 to R2, from R2 to R3, etc. 
>Eventually, R would be deployed to production and R-1 would be 
>retired. 
>
>So, here's my question, so we don't go down the wrong track, 
>can any of the experienced SVN folks comment on the soundness 
>of this approach using SVN? I've been experimenting with it 
>using SVN and it seems to work nicely. Any input would be much 
>appreciated. Thanks
>
>Chuck Irvine
>Embarq
>
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
>For additional commands, e-mail: users-help@subversion.tigris.org
>

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


RE: Release Branches / 3-plus Development Streams

Posted by Ch...@qimonda.com.
I worked for a company that also did this, and with CVS.
In that case, 
- R+1 was never merged to R
- any fix for R+1 also went to R, if applicable
- any fix for R also went to R+1, if applicable

Since SVN and CVS are basically similar revision
control systems as far as ideology is concerned,
I don't see how SVN can change the approach you
have now.

>-----Original Message-----
>From: Irvine, Chuck R [EQ] [mailto:Chuck.R.Irvine@Embarq.com] 
>Sent: Thursday, April 19, 2007 11:05 AM
>To: users@subversion.tigris.org
>Subject: Release Branches / 3-plus Development Streams
>
>This is a repeat of a previous mailing list posting. Pardon 
>the repeat, but no one responded and, since it is a pretty 
>question for us, I thouht that I would try again. Thanks.
>
>.....
>
>We have lately adopted SVN as our standard source code mgt tool. 
>
>Previously, we used CVS and had a standard approach to 
>branching in the context of multiple concurrent release 
>development streams. For two concurrent streams, it was 
>basically what is described in the SVN Book as "Release 
>Branches". Unfortunately, we are often forced to develop for
>3 and 4 releases concurrently. (The SVN book does't comment 
>specifically on this situation.)
>
>For this situation, our branching model using CVS was 
>basically to always have the latest release (R) on the trunk. 
>Whenever we started on release R+1, create a release branch 
>for R and let R+1 development proceed on the trunk. 
>Graphically, it looks like this:
>
>
>             /---------------------
>            / R1
>           /
>          /            /---------------
>         /            /  R2
>        /            /
>-------/------------/-------------------
>R1        R2          R3
>
>We would then merge from R1 to R2, from R2 to R3, etc. 
>Eventually, R would be deployed to production and R-1 would be 
>retired. 
>
>So, here's my question, so we don't go down the wrong track, 
>can any of the experienced SVN folks comment on the soundness 
>of this approach using SVN? I've been experimenting with it 
>using SVN and it seems to work nicely. Any input would be much 
>appreciated. Thanks
>
>Chuck Irvine
>Embarq
>
>
>---------------------------------------------------------------------
>To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
>For additional commands, e-mail: users-help@subversion.tigris.org
>

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org


Release Branches / 3-plus Development Streams

Posted by "Irvine, Chuck R [EQ]" <Ch...@Embarq.com>.
This is a repeat of a previous mailing list posting. Pardon the repeat,
but no one responded and, since it is a pretty question for us, I thouht
that I would try again. Thanks.

.....

We have lately adopted SVN as our standard source code mgt tool. 

Previously, we used CVS and had a standard approach to branching in the
context of multiple concurrent release development streams. For two
concurrent streams, it was basically what is described in the SVN Book
as "Release Branches". Unfortunately, we are often forced to develop for
3 and 4 releases concurrently. (The SVN book does't comment specifically
on this situation.)

For this situation, our branching model using CVS was basically to
always have the latest release (R) on the trunk. Whenever we started on
release R+1, create a release branch for R and let R+1 development
proceed on the trunk. Graphically, it looks like this:


             /---------------------
            / R1
           /
          /            /---------------
         /            /  R2
        /            /
-------/------------/-------------------
R1        R2          R3

We would then merge from R1 to R2, from R2 to R3, etc. Eventually, R
would be deployed to production and R-1 would be retired. 

So, here's my question, so we don't go down the wrong track, can any of
the experienced SVN folks comment on the soundness of this approach
using SVN? I've been experimenting with it using SVN and it seems to
work nicely. Any input would be much appreciated. Thanks

Chuck Irvine
Embarq


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org