Quoting Alex Dubov <[EMAIL PROTECTED]>:
Look, you are not being productive.
I'm sorry you feel that way.
Please, read my letter carefully to the end.
I did not respond to the points I don't have a good handle on (i.e. bucket allocation issues). I'm sure other people that have been involved with APR for much longer know a lot more about it. So, I'll leave it up to them to answer those questions.
I tried responding to the issues that I think are important for the future of DBD API in a way that doesn't say "strings are stupid" or "binary data is stupid". I was trying to be inclusive and to provide pointers for the API that would ensure people don't have to wrestle with wrapping up everything (i.e. keep simple things simple and put more effort into complicated stuff).
The "let's do buckets for the lot" didn't seem like a good idea to me and it still doesn't. It excludes too many valid use cases of the API and is just too complicated for the most simple of use cases. And, if we were to implement binary only passing of parameters, we'd break binary compatibility from the word go.
But, I could be completely wrong on all these points. I'm sure other folks on the list will have their say and whatever turns out good will be put into APR.
I do want to have a good api. I don't know how to design it. I need somebody to do the design of the api for binary data. I outlined the problem in my last message in great details.
Well, I tried suggesting an approach to the API design which I thought would satisfy more than one person, but you didn't find that useful. That's OK.
Apologies if I'm not responding to the questions you really want answered. I'm sure someone on the list eventually will.
-- Bojan
