Lars, as everyone here is painfully aware of by now, I get annoyed by changes to the core API, and core philosophy. And one of the main reasons is that I feel, rightly or wrongly, that these changes can be introduced with the tinyiest of discussions over the shortest period of time. If you go away for a three day weekend, the decision has already been made.
Even though your 20 minute fix is backwards compatible, it still adds to the processing of every db_multirow call. When someone has to wade through that code to trace some bug, in the future, they will have to consider your addition. I'm also not sure why the -unclobber only works if you have a code_block block. Shouldn't it work regardless?
Why not a non-core change? I would suggest a save-restore package, that might take more than 20 minutes to write, but it would not affect the core, and you could re-use the code anywhere it might help.
Pesonally I feel it is the job of the page programmer to work out these naming conflicts. The tcl/adp page are one thing. In fact the entire set of code that is executed in response to a request is one thing. Ignoring this fact will lead to bugs which are hard to trace.
Maybe it has something to do with my name. Whenever I call someone up, even close friends, I introduce myself as 'Tom Jackson'. I realize that the name 'Tom' could mean many things to most of my friends; I want to make sure they understand the correct meaning. Some folks call me up and just start talking, as if they needed no name. I imagine these are Perl programmers of some sort.
The programmer writing a page has the most information and knowledge of what each variable means. Naming them appropriately is the best method of communicating and documenting this information.
Is the logger page you are working on going to become part of the newly released logger package?
