On Tue, Nov 22, 2016 at 10:50 PM, R Smith <rsm...@rsweb.co.za> wrote: > On 2016/11/23 2:08 AM, Scott Hess wrote: >> https://www.sqlite.org/pragma.html has: >> "No error messages are generated if an unknown pragma is issued. >> Unknown pragmas are simply ignored. This means if there is a typo in a >> pragma statement the library does not inform the user of the fact." >> >> I just lost some time due to this, even though I was fully aware of >> it. My code wasn't working, so I instrumented to report errors, and >> gradually dug things deeper and deeper. It wasn't until I was >> verifying statements line-by-line against sqlite3 in a terminal window >> that I saw that I was setting journal_mod rather than journal_mode! >> >> I realize that pragma don't have the compatibility guarantees that >> other syntax has. But that means I actually _would_ want my code to >> start barfing if a PRAGMA stops being supported. Say I'm issuing >> "PRAGMA secure_delete=ON" - presumably it's for a reason, and I'm >> going to want to revisit that reason if it's no longer supported. > > The problem is more the other way round - Backward compatibility is hard > when you introduce a new pragma, and a previous version of SQLite "barfs" > suddenly when it encounters that statement. Recent posts here re-emphasize > the frequency with which old versions are still used out there.... > > Silently ignoring an unrecognized pragma is a great way to be able to > introduce new functionality without worrying that the old will break.
I may be misunderstanding where you're going with that, but my point was that there are some situations where I'm calling a PRAGMA because it's important to some implementation detail. For instance, if your app's schema requires operational foreign keys, then simply ignoring "PRAGMA foreign_keys = ON" is _not_ kosher (I mean, yes, there's nothing SQLite can do to fix being old or compiled that way, but signalling "I can't do that, Dave" would be helpful). How the PRAGMA I am _not_ calling are handled doesn't seem super relevant to that case, because I'm not calling them. Of course, my "PRAGMA pedantic_pragma = ON" call would itself succeed by failing on an older version of SQLite. You have to bootstrap someplace, so you'd have to make the call to turn it on, then do another query to verify that it is on. Which I guess is basically what your PRAGMA should already be doing, though there are subtleties (PRAGMA result patterns aren't consistent across different PRAGMA). -scott _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
