DSQLite a libraryri using sqlite

[Steve Teale]

> Again, Zero coordination in Phobos development. -You are working on
> SQLite,
> -Steve is working on MySQL / and more important .. on std.database
> -Piotr is working on PostgreSQL,
> 
> Steve started his development based a the MySQL CAPI, works almost fine
> until - Ultra-nervous License obsessed people bashed that approach...
> Sad.
> //Beside what was the CAPI manifesto good for ?.
> 
> A fee weeks ago I wrote a mail to Steve, in which I mention that a
> Top-Down development would make more sense . Means Interfaces/ Design
> first.. detail implementation later.
> It is indeed very frustrating but I have to say : Sorry Steve, IMHO you
> don't see the big picture. std.database will fail.
> 
> _SQLite_
> Now you came up with a higher level wrapper around sqlite, which is for
> sure not compatible to whatever Steve or Piotr have in mind. --
> .....I am afraid that Phobos will take another 3 years to consolidate.
> std.database is just another indication for that.. my 2 Euro cents   (
> pretty weak nowadays.

Bjoern,

There was little consensus in the discussion about the higher-level 
std.database, so I feel that I need to know more about what is possible 
before we start on what may well be a lowest common denominator approach 
that will suit nobody.

I've looked at Piotr's stuff, and I can see more similarities than 
differences in the end result of what he is doing and what I am doing.

There is at least one other person I've been in contact with who is 
working on SQLite.

I have done something in this general direction with ODBC in the past, 
and have a fairly clear idea of what can be done there.

Once we have a firm Idea of the set of common facilities that are 
available given licensing restrictions etc I want to provoke the higher 
level discussion again with the focusing element of a background paper 
that summarizes what can be done, and what is actually available as 
building blocks. Then the ORM people and the rest can fight it out and 
we'll see if some sort of consensus can emerge.

In a situation where you are herding cats it might be good to have some 
bait. In this case the prospect of a relatively rapid outcome if people 
could agree on what is needed.

I think that this approach is more likely to work than one where I do 
some sort of high level design and then we just have another round of 
disagreement.

If you have a strong view about what the higher level approach should 
look like, please do a sketch of it. I have no problem if someone else is 
the primary author.

Steve.