I’m using Cursor + Composer with Laravel.
With Laravel, you create database migration files to define the schema of your database.
You typically create these with a CLI command (“PHP artisan make:migration xyz”), which will produce a correctly named but pretty empty file. Part of the naming convention is to prefix the name with a timestamp (eg. 2024_09_21_12345_create_some_table.php)
From there, as a developer, you would populate the file with your DB table schema eg:
public function up(): void
{
Schema::create('groups', function (Blueprint $table) {
$table->id();
$table->string('name')->unique();
$table->timestamps();
});
}
From there, you run php artisan migrate and the database tables will be created in line with the definitions of these files.
When doing this within Cursor: from my description, Composer infers (mostly) correctly the database requirements of an app, and will then provide the commands for creating the file (php artisan make:migration xyz), but will also create migration files itself, but with a name like:
xxxx_xx_xx_create_some_table.php
So then I will have somewhat duplicate files:
- 2024_09_21_12345_create_some_table.php (which is mostly empty/placeholder) and
- xxxx_xx_xx_create_some_table.php (which contains the schema definition).
At this point, I then copy/paste the contents of the Cursor-generated file into the one generated by php artisan make:migration (and delete the Cursor-generated file) - or - I rename the Cursor-generated file to that of the file created by the command (and delete the php artisan generated file).
When we’re handling one file at a time, it’s not so bad, but sometimes 5+ files can be created simultaneously and sometimes when other migration files already exist in the project - this becomes quite tricky to manage.
I would love for Cursor Composer to:
1. Provide the necessary php artisan commands.
2. Wait for my confirmation that I’ve run them.
3. Then modify the newly created files accordingly.
I tried requesting this behavior in my prompt. Cursor paused after providing the commands, but when I instructed it to continue, it created more xxxx_xx_xx type files instead of modifying the existing ones.
Any thoughts on how I can improve this workflow? Thanks in advance for your help!