Closed jussisaurio closed 1 month ago
@pereman2 Wanna review this?
@jussisaurio looks like limit 0 test is busted, can you please fix it?
@jussisaurio looks like limit 0 test is busted, can you please fix it?
Isn't it this test: Test FAILED: 'select count(1) from users where 0;'
? Anyway yeah, it returns NULL
in this and 0
in sqlite, not sure why
Will take a look tomorrow
@jussisaurio @pereman2 I think once we have the test passing, let's merge this to avoid conflicts. But totally agree on moving the call. Codegen optimizations also make sense
The test itself is just wrong. I suspect you got bitten by the fact that our aggregates was bit broken before but got fixed in 93aefa3ee6b8f5a30a5286c28558556387f8ac4c:
The following fixes the problem:
diff --git a/testing/all.test b/testing/all.test
index 0ab665e..252ae03 100755
--- a/testing/all.test
+++ b/testing/all.test
@@ -117,7 +117,7 @@ do_execsql_test where-clause-unary-true {
# not correct? should be 0?
do_execsql_test where-clause-unary-false {
select count(1) from users where 0;
-} {NULL}
+} {0}
do_execsql_test where-clause-no-table-unary-true {
select 1 where 1;
Merging...
Sorry, I didn't realize @Feilkin had expressed interest in implementing this (#80) until I was mostly done... anyway, here's some kind of attempt. My bytecode/assembly-fu is quite basic 🍝