I am proud beyond measure to announce the release of DBD::SQLite 1.25.
This release implements the switch-over from the plain old-style C source to the official SQLite Amalgamated version of the library, which should end the need for DBD::SQLite::Amalgamation and return us to a single driver again.
This release also represents the completion of the "sprint" phase of development, with the entire process from beginning to end taking less than a month.
All primary outstanding issues have now been dealt with fully and robustly.
Unicode is supported properly, we're using the release channel (amalgamation) that SQLite recommends, and we are updated to the most recent version of SQLite. Platform support for 64-bit platforms is (tentatively) in place, the test suite has been modernised to Test::More and greatly increased in coverage, and the bug list in RT has been reduced to only subtle bugs and feature requests. And finally, we've stolen all the useful new features from DBD::SQLite::Amalgamation and merged all patches from external authors that implemented support for new features.
Finally, on a personal level, I think this effort has been the best demonstration yet for the "collective maintenance repository" community model. It shows that a "trust everyone" (letting almost anyone commit to trunk) project structure with centralised version control is still extremely competitive when compared to the new decentralised version control shinyness, especially for small to medium sized projects.
Anti-Flame Note: That said, I freely admit that as the size of a project's code gets bigger, the balance shifts towards distributed version control. If DBD::SQLite was an order of magnitude larger, I'm not sure this would have worked quite as well.
Thank you to everyone that has worked on the module over the last month (in order of appearance in the 281-entry commit log).
Additional thanks to everyone else that has tested, advised, or bug-reported during this process.
I expect to see the release rate slow a bit from here on, pending any urgent bugs appearing in 1.25. Some external testers (companies etc) report they need the module to stay still for long enough that they can do their own more-involved verification testing.
The roadmap from here is to start pushing the XS internals closer to recommended DBI usage, particularly in respect to using newer and more efficient macros, doing warnings properly in the DBI-style, and ferreting out compiler issues on some of the more diverse platforms (HP, 64-bit, etc). I'd also like to see us do some work on adding leak testing.
So I expect the new work to be more detailed and take a little longer, with more intense test scripts.
But regardless, thank you everyone for an extremely successful module handover, and welcome to the next age of SQLite support.