Commented by: @mrotteveel

What is the rationale for this?

Commented by: @dyemanov

It was always possible to use a full pathname for an alias, thus transparently substituting one database location with another (e.g. when aliases are not used but the database should be relocated without touching clients). But it doesn't work if the path contains embedded spaces.

Commented by: Stefan Heymann (stefanheymann)

I think the alias should be an identifier that follows the rules you would otherwise apply to variable names (e.g. first character a letter, then only letters, digits and underscores, maybe dashes).

Commented by: @dyemanov

This will break the feature described above. It may not be widely known and perhaps it even looks "hackery", but I know it's used by some DBAs.

Commented by: @mrotteveel

@dmitry I wonder if that should be considered a feature, or just a hack 'abusing' how aliases work.

Commented by: @AlexPeshkoff

Mark, I do not know was such usage ever documented or not (on the other hand - why document it after the fact that alias is any string?) but that works de-facto since 1.5, therefore I think we should treat changing visible database path using aliases as feature.

Commented by: @mrotteveel

I think the consequence and requirements of this must be fully thought through. For example, it is possible to have a database path like `D:\DB\a=b\database.fdb` / `/var/firebird/a=b/database.fdb`. Defining an alias for that would be hard given the equals (=). So supporting the full syntax requires additional measures (e.g. enclosing such 'keys' in quotes).