Actually all that would happen is a massive number of hidden bugs would be revealed. He would be doing the world a favour.
On Thu, Jan 12, 2017 at 5:13 PM, Jean-Christophe Deschamps <j...@antichoc.net > wrote: > At 15:13 12/01/2017, you wrote: > >> Re: "I read this as a provocative joke." >> >> I didn't read it as just a joke. >> >> The analogy with random fonts, etc. breaks down, I think, because >> randomizing the ordering would be an attempt to *improve* sqlite's >> usability -- not some pedantic punishment. >> > > I read this, as well as Hick previous reply. I'm well aware of the issue, > which is in no way specific to SQLite. > > Yet, providing some new SQLite build (source, amalgamation binaries) > someday where the result order would be willingly random or different from > the current behavior (call it natural or naively expectable or intuitive or > whatelse) will break uncountable uses where the app isn't open to change. > Remember that in many situations SQLite is being used as a loadable > component either because the original code was designed so or because the > language used can't statically link. > > So it could be an improvement for *-future-* SQLite apps, or rather a good > reminder aimed towards developpers, but that would potentially break > gazillions legacy uses or at the very least cause a huge lot of unnecessary > inconveniences. Expect a tsunami of disapprovals. > > If a user has problems with her sqlite output early in the process, leading >> to the discovery of a missing "ORDER BY" clause, the argument is that she >> has been dealt a favor. It's vastly worse for her to encounter a >> mysterious bug when the sqlite version is updated years from now to one >> which (perfectly correctly) returns a different ordering for that same >> query. >> >> Further, Dr. Hipp and his team won't have to deal with howls of "it's >> broken" when such a version is released. >> > > I also have to repeatedly point out in the community where I offer support > that SQL deals with unordered sets and to the consequence, that issuing the > very same SELECT twice in a row could rightfully return results in > different orders when no ORDER BY clause is specified. But I bet such an > uncalled change (as salutary as it may be from a rational point of view) > would result in a long term continuous higher saturation of this list and > other support channels with posts from questionning/angry/disappointed > users. > > > > _______________________________________________ > sqlite-users mailing list > sqlite-users@mailinglists.sqlite.org > http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users > _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
