Don't get me wrong, I'd say that's a very good approach too. A lot of my personal repos are private and synced through personal infrastructure so I don't always bother with those kind of files.
I habitually start with 'touch .gitignore' and then committing that.
Since an empty .gitignore doesn't actually do anything and has never been an issue to rebase onto, it works out fine for me.
I should probably, arguably, switch to using --allow-empty but my current workflow is long since baked into my fingers and has yet to cause a problem for me or anybody else, so if I switch it'll be for reasons of pure pedantry (which me being me may eventually be quite sufficient motivation, but hasn't just yet).
Since an empty .gitignore doesn't actually do anything and has never been an issue to rebase onto, it works out fine for me.
I should probably, arguably, switch to using --allow-empty but my current workflow is long since baked into my fingers and has yet to cause a problem for me or anybody else, so if I switch it'll be for reasons of pure pedantry (which me being me may eventually be quite sufficient motivation, but hasn't just yet).