CI: Clippy, MySQL & Postgres

[billy1624]

No description provided.

[billy1624]

Discovery behaviour difference on MySQL vs MariaDB. Creating the same table but schema discovery results are different. @tyt2y3

MySQL, https://github.com/SeaQL/sea-schema/runs/3974821329?check_suite_focus=true#step:10:1164

Diff < left / right > :
<"CREATE TABLE `bakery` ( `id` int(8) NOT NULL AUTO_INCREMENT, `name` varchar(255), `profit_margin` double, PRIMARY KEY (`id`) ) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_general_ci"
>"CREATE TABLE `bakery` ( `id` int NOT NULL AUTO_INCREMENT, `name` varchar(255), `profit_margin` double, PRIMARY KEY (`id`) ) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_general_ci"

MariaDB, https://github.com/SeaQL/sea-schema/runs/3974821107?check_suite_focus=true#step:10:1298

Diff < left / right > :
<"CREATE TABLE `bakery` ( `id` int(8) NOT NULL AUTO_INCREMENT, `name` varchar(255), `profit_margin` double, PRIMARY KEY (`id`) ) ENGINE=InnoDB DEFAULT CHARSET=utf8 COLLATE=utf8_general_ci"
>"CREATE TABLE `bakery` ( `id` int(8) NOT NULL AUTO_INCREMENT, `name` varchar(255) DEFAULT NULL, `profit_margin` double DEFAULT NULL, PRIMARY KEY (`id`) ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb3 COLLATE=utf8mb3_general_ci"

[tyt2y3]

Both are semantically equivalent. It's hard to be verbatim exact.

[billy1624]

Perhaps we can skip the mysql/live checks for now

[tyt2y3]

Well we actually need a looser way to compare two statements, like ignoring the integer widths and 'default null'.

But Postgres is all fine?

[billy1624]

Well we actually need a looser way to compare two statements, like ignoring the integer widths and 'default null'.

But Postgres is all fine?

All fine

[billy1624]

Hey @tyt2y3, seems that for MySQL 8 the information schema changed? It fail to parse the column length of integer column.

Diff < left / right > :
<"CREATE TABLE `bakery` ( `id` int(255) NOT NULL AUTO_INCREMENT, `name` varchar(255), `profit_margin` double, PRIMARY KEY (`id`) ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci"
>"CREATE TABLE `bakery` ( `id` int NOT NULL AUTO_INCREMENT, `name` varchar(255), `profit_margin` double, PRIMARY KEY (`id`) ) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_general_ci"

GitHub Actions

• MySQL 8: https://github.com/SeaQL/sea-schema/runs/4103850673
• MySQL 5.7: https://github.com/SeaQL/sea-schema/runs/4103850846

[tyt2y3]

I think it was deprecated in 8.
I am thinking to override the default impl of ColumnDef such that the width of integer columns are ignored.
Then it can handle the previous failed case in test suite too.

[tyt2y3]

Okay a better idea is to have a normalise() method to strip out the irrelevant stuff so they will be equivalent.

[billy1624]

Okay a better idea is to have a normalise() method to strip out the irrelevant stuff so they will be equivalent.

Strip the asserting String directly?

[billy1624]

Ready for review & merge @tyt2y3