On 17/01/2017 10:14, Dimitry Sibiryakov wrote: > 17.01.2017 13:03, Adriano dos Santos Fernandes wrote: >> If any other solution allows "valid state (initial) -> invalid state >> (uncommitted) -> valid state (committed)" more DML changes in the same >> objects, it's ok for me. Personally, I think the concepts does not match. > But what exactly "invalid state" means? I can't imagine anything but > violation of > "foreign key" on table rdb$dependencies. If Firebird ever is going to have > deferred > constraints, allowing some metadata to be inconsistent during transaction and > their > validation on commit should be relatively simple using them, no? > > Everything that causes user to currently use tools that "recompile" (recreate or alter to empty bodies and alter again).
It means, allows the user to alter objects in incompatible way with its currently dependencies. Changing a function return type from integer to string, when dependencies sum the call to that function, for example. We have the INVALID state for PSQL currently, but I'm talking at general (using in views, computed, etc). If in the transaction end everything is compatible, allow the change. Adriano ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot Firebird-Devel mailing list, web interface at https://lists.sourceforge.net/lists/listinfo/firebird-devel
