On 5/15/07, Andrew Scott <[EMAIL PROTECTED]> wrote:
> Deploy? If you mean deploy to production, thats always trunk.

That's actually a really bad process in general -- b/c then you have
to keep track of revision numbers to be able to revert to a previous
release. The standard advice is to tag the trunk (eg REL-20070515 or
REL-1.2.3) and then deploy the tagged release, which means exporting
from the tag instead of trunk. That gives you a MUCH better history
over time.

Deploying from /trunk directly is generally a good way to handle
continuous integration, but I recommend not doing that for production.
Tagging gives the added benefit of repeatability without needing to
track revision numbers, though clearly nothing stops you from doing it
this way :)

> Or do you mean deploy from your code to the repository, that's commit.

>
> On 5/15/07, AJ Mercer <[EMAIL PROTECTED]> wrote:
> >
> > OK, irrespective of whether the trunk or branches are used
> > how do you selectively chose what you want to deploy?
> >
> > If I have 5 bug fixes on the go bug001 - bug005,
> > and bug004 is signed off
> > how would I use subversion to know which files need to be updated on
> > production (without any of the work on bugs 1,2,3,5 going over)?
> >
> > On 5/15/07, Andrew Scott <[EMAIL PROTECTED]> wrote:
> > >
> > > AJ,
> > >
> > > Welcome to the world of many ways to skin a cat.
> > >
> > > We do it the same way I outlined a few times now, but others have
> > diffeent
> > > ways to do it.
> > >
> > > When working on something that is either new or a bug, the code is
> > looked
> > > at
> > > and fixed now until it is fixed, tested with unit tests and the
> > > development
> > > tested we do not commit it back to SVN.
> > >
> > > Thats our procedure, but as I said earlier, we also sync the changes to
> > > make
> > > sure we don't need to merge any new code, and then when its tested,
> > merged
> > > on our code with the repository we then commit it.
> > >
> > > Thats our method, and others have many other ways of doing it.
> > >
> > >
> > >
> > > On 5/15/07, AJ Mercer <[EMAIL PROTECTED]> wrote:
> > > >
> > > > If by bug you mean typo or something minor - then making the change to
> > > the
> > > > trunk would seem fine.
> > > >
> > > > I am think more along the lines where 2 or more pages need to be
> > > modified
> > > > and may span a number of days (elapsed time)
> > > >
> > > >
> > > > When it comes to committing , would it be fair to say:-
> > > > trunk: only commit logical units of work that will not break the
> > system
> > > > branch : knock yourself out
> > > >
> > > >
> > > > With so many different views and recommendation it is no wonder I am
> > > > getting
> > > > confused :-}
> > > >
> > > > On 5/15/07, Andrew Scott <[EMAIL PROTECTED]> wrote:
> > > > >
> > > > > Seriously,
> > > > >
> > > > > I would not create a branch for a bug, just to have it merged back
> > to
> > > > the
> > > > > trunk again.
> > > > >
> > > > > You might as well apply the fix directly on the trunk...
> > > > >
> > > > >
> > > > >
> > > > > On 5/15/07, John Paul Ashenfelter <[EMAIL PROTECTED]>
> > > > wrote:
> > > > > >
> > > > > > On 5/14/07, AJ Mercer <[EMAIL PROTECTED]> wrote:
> > > > > > > I am still not quite sure how to go about deploying mods from
> > > > > > subversion.
> > > > > > > This is what I am trying to achieve:
> > > > > > >
> > > > > > > Each developer has a local development environment.
> > > > > > > We have a job tracking systems (jira) that bugs and requests for
> > > new
> > > > > > > features are entered into.
> > > > > > >
> > > > > > > I will work on a bug and once I am happy with the fix, testing
> > in
> > > my
> > > > > dev
> > > > > > > environment, I will put those changes onto the test server.
> > > > > >
> > > > > > We're missing some details here about the process around
> > subversion
> > > so
> > > > > > it's hard to directly answer the question, but one very common
> > > pattern
> > > > > > is to create a bug-specific branch (usually off of the release
> > > branch)
> > > > > > and fix the bug there. The full process would work something like
> > > > > > this:
> > > > > >
> > > > > > * assume there's an existing branch for the release,
> > > > /repo/branches/RB-
> > > > > 1.0
> > > > > > * copy the release branch /repo/branches/RB-1.0 to a bugfix branch
> > > > > > /repo/branches/BUG-3456
> > > > > > * tag the beginning of the bugfix branch /repo/tags/PRE-3456
> > > > > > * work on bug in the bugfix branch committing as necessary
> > > > > > * tag as necessary on bugfix branch, eg /repo/tags/QA-3456,
> > > > > > /repo/tags/QA-3456-RC2, etc
> > > > > > * when the bug's signed off on, tag the end of the branch
> > > > > > /repo/tags/POST-3456
> > > > > > * back in the release branch, /repo/branches/RB-1.0, you can merge
> > > > > > /tags/PRE-3456 and /tags/POST-3456 to get the release updated
> > > > > > * roll out a new release (tagged and/or branched of course) or
> > > > > > generate a patchfile depending on your process
> > > > > > * check and see if the merge needs to go back into trunk/etc as
> > > > > > appropriate
> > > > > >
> > > > > > Obviously this isn't strictly necessary for a typo or a very small
> > > > > > change, though at a minimum using the tags for PRE and POST will
> > > make
> > > > > > your life easier.
> > > > > >
> > > > > > As an aside, Subversion revision numbers are not in short supply
> > --
> > > > > > there's no need to conserve them, and since copies in SVN are
> > > > > > efficient, there's no worry about increasing the size of the repo
> > > > > > every time you branch (like in BitKeeper and some other source
> > > control
> > > > > > systems).
> > > > > >
> > > > > > We'll cover this stuff in my CFUnited 2007 pre conference class.
> > At
> > > > > > worst, if you're reading this and figuring you need to know more
> > > about
> > > > > > Subversion, you should check out the Pragmatic Version Control
> > using
> > > > > > Subversion book from the Pragmatic Programmers and save yourself a
> > > lot
> > > > > > of effort in learning SVN.
> > > > > >
> > > > > > > After user testing, it may come back for further work, or be
> > > signed
> > > > > off.
> > > > > > > If signed off, it is put onto the production server.
> > > > > > >
> > > > > > > As you can imaging, there can be any number of bugs been fixed
> > at
> > > > any
> > > > > > point
> > > > > > > in time by multiple developers.
> > > > > > > The testing can get signed off in any order and may be there for
> > a
> > > > > > couple of
> > > > > > > weeks.
> > > > > > >
> > > > > > > What I would like to be able to do is get the files for a
> > > particular
> > > > > bug
> > > > > > fix
> > > > > > > and from those file, merge the modifications into test and then
> > > > > > production.
> > > > > > >
> > > > > > > General Notes:
> > > > > > >
> > > > > > >  - One file may have multiple bug fixes / enhancements
> > > > > > >
> > > > > > >  - if a modification is to be backed out, and other modification
> > > > since
> > > > > > then
> > > > > > > will need to stay.
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
>
> 

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~|
Upgrade to Adobe ColdFusion MX7
Experience Flex 2 & MX7 integration & create powerful cross-platform RIAs
http://www.adobe.com/products/coldfusion/flex2/?sdid=RVJQ 

Archive: 
http://www.houseoffusion.com/groups/CF-Talk/message.cfm/messageid:278149
Subscription: http://www.houseoffusion.com/groups/CF-Talk/subscribe.cfm
Unsubscribe: http://www.houseoffusion.com/cf_lists/unsubscribe.cfm?user=89.70.4

Reply via email to