vtexplain doesn't handle float columns

[slanning]

vtexplain with a simple table succeeds:

$ vtexplain -schema 'CREATE TABLE i (i INT NOT NULL,PRIMARY KEY(i))ENGINE=InnoDB'
-sql 'select 1 from i' -vschema '
{"s":{
"sharded":false,
"vindexes":{"hash":{"type":"hash"}},
"tables":{"i":{"column_vindexes":[{"column":"i","name":"hash"}]}}
}}'

select 1 from i

1 s/-: select 1 from i limit 10001

---

If I add only an "f float" column, it says it's unimplemented:

$ vtexplain -schema 'CREATE TABLE i (i INT NOT NULL,f float,PRIMARY KEY(i))ENGINE=InnoDB'
-sql 'select 1 from i' -vschema '
{"s":{
"sharded":false,
"vindexes":{"hash":{"type":"hash"}},
"tables":{"i":{"column_vindexes":[{"column":"i","name":"hash"}]}}
}}'
E1019 13:32:41.561362 18012 exit.go:89] panic: unimplemented type float
/usr/lib64/go/src/runtime/asm_amd64.s:509 (0x45c7fa)
call32: CALLFN(·call32, 32)
/usr/lib64/go/src/runtime/panic.go:491 (0x42d992)
gopanic: reflectcall(nil, unsafe.Pointer(d.fn), deferArgs(d), uint32(d.siz), uint32(d.siz))
$HOME/go/src/github.com/youtube/vitess/go/vt/sqlparser/ast.go:939 (0x81a7b9)
com/youtube/vitess/go/vt/sqlparser.(*ColumnType).SQLType: panic("unimplemented type " + ct.Type)
$HOME/go/src/github.com/youtube/vitess/go/vt/vtexplain/vtexplain_vttablet.go:266 (0xa654a9)
com/youtube/vitess/go/vt/vtexplain.initTabletEnvironment: Type: col.Type.SQLType(),
$HOME/go/src/github.com/youtube/vitess/go/vt/vtexplain/vtexplain.go:134 (0xa5ee0f)
com/youtube/vitess/go/vt/vtexplain.Init: err = initTabletEnvironment(parsedDDLs, opts)
$HOME/go/src/github.com/youtube/vitess/go/cmd/vtexplain/vtexplain.go:173 (0xa6b81d)
parseAndRun: err = vtexplain.Init(vschema, schema, opts)
$HOME/go/src/github.com/youtube/vitess/go/cmd/vtexplain/vtexplain.go:140 (0xa6b2c0)
main: err := parseAndRun()
/usr/lib64/go/src/runtime/proc.go:185 (0x42f79c)
main: fn()
/usr/lib64/go/src/runtime/asm_amd64.s:2337 (0x45f060)
goexit: BYTE $0x90 // NOP

That line in ast.go seems to want to handle float types, at first glance at least.

[demmer]

Oops -- this was due to a simple typo due where I used the FLOAT instead of FLOAT_TYPE token id.

I just pushed a PR that should fix this.

[slanning]

I just came to the same conclusion. Thanks for the fast fix!