Re: Source Control for DB objects

  • From: Subodh Deshpande <deshpande.subodh@xxxxxxxxx>
  • To: jeremy.schneider@xxxxxxxxxxxxxx
  • Date: Thu, 4 Aug 2011 20:56:14 +0530

See look jermy,
Dba's recent statement, "if its not broken do not fix it"; i think this is
justifiable if and only if there is no proper roll out process followed or
very poor documentation available. A broken code is harmful for data as well
as for database and for this dba is held responsible and thats why such
comment may occure.

About source control system, i would like to have name of the developer,qa
person and also approver's name for every approved and implemented change.
And this is not sufficient this is minimum, the change should either
describe what that change is in detail or it should point to an existing
document that describe this change. By change in detail i mean it should
inform about schema, schema objects i.e. Affected tables, columns,
procedures, functions,packages,triggers, indexes, change in business logic
and hence affected interfaces such as forms or reports.
This is why i think one should design process to maintain change in source
control and which practices need to be followed so that SRS will be
fulfilled within the limit of SLA, cause finally you or atleast answerable
to yourself.

On 4 Aug 2011 20:01, "Jeremy Schneider" <jeremy.schneider@xxxxxxxxxxxxxx>
wrote:

Personally I'm a big fan of git for version control... and I do have a bit
of a personal soap-box about how DBAs are really developers but act like
they're not.  (We don't ever seem to follow many well-thought out practices
about code development... even though we often manage thousands of lines of
code in shell scripts/SQL and change lots of configuration & schema related
things spontaneously without any kind of record or reproducibility...)

At two recent large orgs where I worked, they were still using CVS for the
DBA group.  This was not because CVS is great, but just because it was there
and the DBA's didn't care much about recent advancements in development
practices.  (Very recent quote from one DBA: "if it's not broke, then don't
fix it.")  Subversion does seem to be pretty popular on the whole - and it
is integrated with quite a few tools.  Although I think that you can
increasingly find plugins for the more recent source control packages too
(like git or mercurial)... and of course there are the commercial
alternatives too, like clearcase (used at another large org where I
worked).  If I remember right, Oracle's OSS website runs on subversion and
the linux kernel uses git.

I'm interested to follow this thread - and hear what other people are
using.  In particular, I'd also be curious what development tools you use
(Oracle's SQL Developer? Toad?) and if your tools integrate with your source
control system for PL/SQL scripts.  I'd also be curious to hear what you put
into change control - just PL/SQL code, or also system configuration/schema
updates?  How about shell scripts?

-Jeremy




On Wed, Aug 3, 2011 at 1:42 PM, <Joel.Patterson@xxxxxxxxxxx> wrote:
>
> It used to be that Source...
-- 
http://www.ardentperf.com
+1 312-725-9249

Jeremy Schneider
Chicago

Other related posts: