On 12. des. 2004, at 21.39, Simon Flack wrote:
> Dave Howorth wrote:
>> Simon Flack wrote:
>>> Yes, I like that. One issue is that the table2class() differs from
>>> Class::DBI::Loader. E.g.
>> I prefer AutoLoader's rule, but I guess one issue might be backwards
>> compatibility.
>>> table: hello-world
>>> CDBI::AutoLoader: HelloWorld
>>> CDBI::Loader: Hello-world # this is an invalid package name
>> So there's no backwards-compatibility issue there then :)
>>> table: hello__world
>>> CDBI::AutoLoader: HelloWorld
>>> CDBI::Loader: Hello_World
>> That will either work, if the user hasn't actually written any code
>> for the table class, or it will cause a compile error that's easy to
>> fix in the more usual case that there is code for an untaint or a
>> relationship or something.
>> The most worrying case I can think of is if somebody has one table
>> called hello__world and another called hello_world. But then I think
>> they deserve everything that happens to them :) And something would
>> probably break.
>> The other issue will be Class::DBI::Loader::Relationship.
>
> That's a tricky one. I think our best bet is proposing a patch to
> Class::DBI::Loader.
I think so too. Replacing Class::DBI::Loader at this stage sounds like
a very bad idea.
Marcus
This archive was generated by hypermail 2.1.3 : Thu Feb 24 2005 - 22:25:57 GMT