On 4/24/2011 3:45 PM, Alex Khmara wrote:
On Sun, 10 Apr 2011 01:19:43 +0300, Jonathan M Davis
<jmdavisp...@gmx.com> wrote:
If it's just the bindings, then we probably don't need an actual review
process on the newsgroup - though it'll obviously be reviewed on
github before
being merged in. However, I think that this sort of binding raises a
potential
problem that we need to make sure we're willing to deal with and know
how to
deal with, and that's library versioning.
How stable is SQLite's API? If they put out SQLite 3.7.6, will any of the
bindings then be invalid? I'm guessing not, since that's likely just a
bug fix
release, but what about 3.8.x? I'd fully expect that SQLite 4 would
have major
changes were it ever released. Does that mean that Phobos then
requires the
user to have a specific version of SQLite installed if they intend to
use any
of the SQLite bindings? If that's the case, then it could make Phobos
unusable
on some systems when SQLite gets upgraded.
If SQLite's API is stable enough (and I think that there's likely a good
chance that it is), then including bindings for it in Phobos makes
good sense,
and it's a step towards including a D-based database solution in
Phobos. But
if SQLite's API changes much at all, then we could have versioning
issues. If,
on the other hand, SQLite bindings were in a separate library from Phobos
(even an official one), then we could have different versions of the
bindings
(and different versions of the eventual D database solution) available by
having different versions of that library, thereby allowing folks to
use the
version of SQLite that's appropriate for their system. But you can't
do that
in Phobos itself.
This is a problem posed by any C bindings that we might want to put into
Phobos. And as long as the APIs are stable enough, it shouldn't be a
problem.
But any project that's likely to have API changes which would affect the
bindings could create versioning issues with Phobos' dependencies. So,
the
question is how we want to deal with that and whether SQLite is stable
enough
that such versioning issues are unlikely to be a problem short of
SQLite 4.
- Jonathan M Davis
SQLite project pay high attention to compatibility. AFAIK, all changes
except
first digit are upward compatible - you can take bindings for 3.5.0 and
link
wih 3.7.6 - all will work as expected, You just will not see new API.
P.S. I updated bindings version to 3.7.6.2 and fixed one bug (see github).
Thank you for your hard work, Alex!
I consider it a no-brainer to include just the bindings and SQLite C
code in etc right now. SQLite is arguably the most widely used
database, is easy to distribute and is public domain. It would be great
if this infrastructure was "just there" for whoever wants to build on it.
As far as the wrapper, I would love a simple SQLite wrapper in Phobos,
since SQLite is the only DB I personally care about. However, it seems
others in the community are interested in a more general SQL DB wrapper
that can be used with a variety of backends. Now that no GSoC database
project has been accepted, we need to consider other options for getting
this done. I understand that there are a lot of independent attempts,
but I don't know the status of them or which ones, if any, are targeting
eventual inclusion in Phobos.
