+1. Now get ready: "With great power comes great responsibility..."
It looks good, but so far it's just variable replacement. How does one go about modifying a specific table/column definition for a database? Would I have to parse the control sql myself? I only see a one liner for parsing/var replacement. But even then, my worry is that a developer can drift away from the control sql enough to cause some damage. Didn't someone mention Torque or some other Apache project to generate db specific sql from templates? However, I bet it has many limitations and it won't make it easy to do the actual db specific customizations we need to work around the idiosyncrasies. Anyways, remember that even after all that rambling, I'm still +1 (because it's much better than what we had before). Elias On 12/5/05, Dave Johnson <[EMAIL PROTECTED]> wrote: > Fixing that PostgreSQL upgrade problem was extremely painful with our > text-replace based database script generation technique. Take a look at > the stuff I committed yesterday if you don't believe me. So after I > finished the work, I replaced the text-replace stuff with a much more > powerful and flexible generator: Velocity (driven by the Texen task). > I've got code ready to commit. The proposal is here: > > http://rollerweblogger.org/wiki/Wiki.jsp? > page=Proposal_VelocityGeneratedDatabaseScripts > > DIFF says the scripts I'm generating with the new technique are > identical to the old scripts. I'd like to commit this stuff today in > the trunk. No matter how you feel about generating database scripts > (I'm for it), you have to admit that this is a big improvement over > text-replace. > > Any objections to this proposal? > > - Dave > > > > >
