Weird result when using query instead of prep_exec

[wqfish]

When debugging an issue in my application that uses mysql_async, I ran into a problem where I get different result by calling query and prep_exec.

I have reduced the program to the following to reproduce the issue:

Cargo.toml: https://gist.github.com/wqfish/7dddb18da14682556e5c33639eeb7b20
main.rs: https://gist.github.com/wqfish/c4a69477b90880c2c5731ce8f90d4747

The content of the db is pretty simple. The db mydb has a single table my_table that has two rows in it: https://gist.github.com/wqfish/21125e61fc305ba2958083fc795dbef7

The problem

With this program I can successfully load the two rows from mysql. However, if I comment out line 58 in main.rs and use line 57 instead, I only get one of the two rows. So changing prep_exec to query caused the result to be different.

Does this feel like a bug or I'm missing something?

[wqfish]

Another way to demonstrate the problem is the following:

Let's just use query instead of prep_exec, so I can not get both rows with the above program. However if I make the following change to the code

diff --git a/main.rs b/main.rs
index 5c86d32..8c3ee75 100644
--- a/main.rs
+++ b/main.rs
@@ -53,7 +53,7 @@ async fn main() -> Result<(), mysql_async::error::Error> {
 
     let conn = conn.drop_query(r"USE mydb").await.unwrap();
 
-    let q = "SELECT b, c, d, e FROM my_table";
+    let q = "SELECT a, b, c, d, e FROM my_table";
     let result = conn.query(q).await.unwrap();
     // let result = conn.prep_exec(q, ()).await.unwrap();
 
@@ -70,7 +70,7 @@ async fn main() -> Result<(), mysql_async::error::Error> {
 
     let (_ /* conn */, loaded_structs) = result
         .map_and_drop(|row| {
-            let (b, c, d, e): (_, _, _, _) = mysql_async::from_row(row);
+            let (a, b, c, d, e): (u64, _, _, _, _) = mysql_async::from_row(row);
             MyStruct { b, c, d, e }
         })
         .await

(Basically I select one more column from the DB.) Then I get both rows.

[wqfish]

I spent some time debugging and tracked down to this code where we parse the packet to determine if there are more results. It seems that the code returned None when it's supposed to return Some. I don't really know the format of the packets to debug further.

[blackbeam]

Hi. Thanks for report!

Could you please test this using the latest commit on master?

[wqfish]

@blackbeam Yeah it seems to be working now on latest master. Thanks for the quick fix!

[wqfish]

Before I close this issue, do you mind explaining why blackbeam/rust_mysql_common#17 fixes this? I don't know much about mysql and all I can tell is that I'm not using floats in my code here.

[blackbeam]

Before I close this issue, do you mind explaining why blackbeam/rust_mysql_common#17 fixes this?

Oh. No, your problem is fixed by 8dd31c1.

The blackbeam/rust_mysql_common#17 is just another breaking change that I wanted to include into mysql_common v0.21.0 release.

[wqfish]

Oh that makes sense. Thanks!