Conditionals control flow should flow with migrations

Describe the bug The check {if index exists} working in reverted way.

To Reproduce The following statement:

if (Schema.Table("core_users").Index("ix_core_users_discriminator").Exists())
      Delete.Index("ix_core_users_discriminator").OnTable("core_users").OnColumn("discriminator");

throws an error: ORA-01418: specified index does not exist Sql: DROP INDEX ix_core_users_discriminator

Expected behavior It should not attempt to delete the index

Information (please complete the following information):

• OS: Windows 10
• Platform .NET Core 3.1.x
• FluentMigrator version 3.2.6
• FluentMigrator runner
• Database Management System: Oracle
• Database Management System Version: “Oracle 18xe”

Additional context If I revert the statement to:

if (!Schema.Table("core_users").Index("ix_core_users_discriminator").Exists())
      Delete.Index("ix_core_users_discriminator").OnTable("core_users").OnColumn("discriminator");

Then it works with no issue. See the “!”.

[jzabroski]

But I can see now those exists checks need to be used carefully assuming they are executed at the beginnig of the script and that the index name must be unique across the scheme (not table like in mssql).

I did create a batch actions ticket recently to make some of this easier to use: #1221

Separately, I had a separate idea for allowing monadic-like F#-like computation expressions in FluentMigrator. See: #987 The idea is that one could use something like LINQ IEnumerable interface and the LINQ syntax to define a monadic-like computation that builds up a list of side effects (aka migrations) and then executes them. While you might not think of an existence check as being a side effect, it does require querying the “outside world” and then taking a destructive action, so it is a side effect. But, in my judgment, there are higher priority features and I don’t know if most users would understand or like the idea vs. just fixing bugs and adding more common features like tablespaces/partitions and other things.

[jzabroski]

Thanks. I probably won’t look any further today but I have a checklist of things to investigate. Ideally I’d like to include a fix (or respond with correct usage syntax) before 3.2.7 is published, which I have a hard cut-off goal of Sunday night for 3.2.7 fixes.