On 2014/12/08 15:58, Gwendal Roué wrote:
I'm new to this mailing list, and I won't try to push my opinion, which is : yes this is a bug, and this bug could be fixed without introducing any regression (since fixing it would cause failing code to suddenly run, and this has never been a compatibility issue). Thank you all for your support and explanations. The root cause has been found, and lies in the constraint checking algorithm of sqlite. I have been able to find a work around that is good enough for me. Now the subject deserves a rest, until, maybe, someday, one sqlite maintainer who his not attached to the constraint-checking algorithm fixes it. Have a nice day, Gwendal Roué
Your new-ness is irrelevant, if you have a worthy argument it deserves being heard. To that end, let me just clarify that nobody was saying the idea of deferring the constraint checking is invalid or ludicrous (at least I had no such intention) and you make a valid point, especially since most other DB engines do work as you suggest - and this will be fixed in SQLite4 I believe, where backward-compatibility is not an issue.
The reason I (and others) will say it isn't a bug is because it isn't working different than is intended, or more specifically, than is documented. It works exactly like described - whether you or I agree with that paradigm or not is up to discussion but does not make it a "bug" as long as it works as described.
I hope the work-around you found works great! _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
