On Fri, Nov 30, 2018 at 3:00 AM J. King <jk...@jkingweb.ca> wrote: > On 2018-11-29 20:56:13, "Richard Hipp" <d...@sqlite.org> wrote: > >On 11/29/18, J. King <jk...@jkingweb.ca> wrote: > >> Is it possible to make SQLite fail like PostgreSQL does? > > > >That is possible in theory, but how many of the millions of existing > >applications would that break? > > > > I was wondering more if it's -currently- possible via some switch I'm > not aware of. As an enhancement obviously I wouldn't expect it as a > default (not before SQLite 4, anyway), but as a debugging feature like > reverse_unordered_selects, it might help to uncover millions of bugs. ;) >
+1. There are many SQLite gotchas for backward compatibility reasons. Regularly those come back up on this list, because someone's been bitten by it. Gotcha is explained, BC is invoked for not fixing it, pragma is suggested on an opt-in basis to have a "stricter and safer" SQLite, then nothing happens. Some gotchas require file-format changes, so are mostly off-limit completely. Although myself I'd welcome a new and non-BC format, getting rid of all legacy. DRH has expressed several times he's not willing to go there, that's just life :) But when the gotcha is pure runtime, it's less easy to accept the status quo, I have to be honest. Especially since Richard often does not take the time to explain his rational for not doing them. It's probably obvious to him, from cost of implementation, or maintenance/testing (to maintain 100% line/branch coverage), or even from disagreeing with the proposed features for such or such reasons. But it's not always obvious to me for sure, and I confess to difficulty accepting the silence. I realise it's a bit unfair to blame Richard when he gives away SQLite, such a wonderful library, and that he's super busy, creating Fossil, and forums, and SMTP servers, and what-not we are not even aware of, for his commercial clients. Yet I still wish there was a bit more transparency and discussions around SQLite, with the "real" stakeholders of SQLite, i.e. DRH and his small dev team. We do discuss things at length here, but the only authoritative voices are Richard and Dan, no offence to all the other great contributors on this list, and Richard and Dan are often making themselves scarce in these threads. I just wish they got involved more, and shared more insights and rationals. Xmas is around the corner, so it's that time of the year I guess :). I'll shut up now. Sorry for the noise. --DD _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
