Le 27/09/2013 23:23, Jonas Sicking a écrit :
On Wed, Sep 25, 2013 at 3:39 PM, Michael Fitchett
<michael.fitch...@spotsync.com> wrote:
Dear Members of the W3C Consortium::
Regarding: Making the W3C Web SQL Database Specification Active
I would like to request that you make the W3C Web SQL Database
specification active again. The Web SQL Database Specification enables
developers to build web-based applications that can store, retrieve,
manipulate and query against data on the client machine. This technology is
similar to SQLite, Microsoft SQL Server, MySQL, etc. Web SQL combined with
Manifest enable developers to build web-based applications that work while
offline.
The Web SQL Database specification was on the W3C Recommendation track, but
the specification was stopped because Mozilla and Microsoft did not want to
implement a specification that lacked proper SQL definition. I know there is
a need for both a NoSQL and SQL solution. The two specifications (Web SQL
Database and Indexed Database API) that exist to date are acceptable..
However, as stated above, the problem is the lack of definition for SQL.
Since lack of definition is the issue, I would like to recommend a remedy.
I know SQL experts and great documentation writers who I would gladly hire
to further define the Web SQL Database specification and fill in the missing
SQL definition. Is this something that would be possible to help revive the
specification and get the remaining vendors on board?
The minimum requirements for bringing back WebSQL, or any other
SQL-based web spec is IMHO:
1. A specification for the SQL dialect being proposed.
2. *Two* independent, production quality, database implementations
being willing to implement exactly that SQL dialect. Not a subset of
it, and not a superset of it.
3. The two independent implementations need to have roughly the same
performance characteristics. I.e. it's not ok for an implementation to
generate correct results, but do it so slowly that it's in practice
unusable.
I'd like to add another requirement which is having a significant
advantage over IndexedDB. If web devs want SQL, they can have it on top
of IndexedDB in the form of an open source library (I'm willing to be it
already exists). They don't need to wait for a standard to emerge, nor
for browsers to consistently implement it.
If they really want a spec, they can create a W3C community group (or a
Github repo). We don't need browsers to do all the work for us!
David
