Both my my first and second points were in regards to your custom (theoretical if you prefer) version. I was explaining what your changes would mean if it was really worded that way to cover all the bases.
And again, you still don't understand that the properties of the weapon you're using or any factor beyond the fact that the SPELL is letting make the ATTACK doesn't matter for determining if the attack bypasses or not. It's like you're glossing over the fact that the errata is talking about SOURCES of attacks. Spells, Magical Items, Magical Sources. Not weapon properties, attack effects, damage dice, damage type.
You're completely stuck on item properties and attack effects, damage type and so on. If you can't see beyond this point nothing I say will get through to you. You're even now changing the words in the spell text to justify your point of view. The 1d6 nonmagical slashing damage that you just posted IS IRRELEVANT. IT doesn't matter. All that matters is the ATTACK is granted by the SPELL which is the SOURCE. You can substitute whatever the hell you want into the spell text, it doesn't matter. It's a spell letting you make an attack ERGO it bypasses. lol
This is a specific beats general rules system. Something has to specifically call out that it allows a game element to bypass the general rules. Simply saying that an attack does 1d6 nonmagical slashing damage doesn't change any of the facts I've just explained to you.
I'm going to quote the errata's exact wording because the debate appears to be drifting from it.
"A magical attack is an attack delivered by a spell, a magic item, or another magical source"
Delivered, as mentioned a couple pages back, can mean one of two things. It can mean (1) that the spell, magic item, or magical source conveys harm to a creature, or (2) that it produces the attack that then harms a creature. If delivered refers to the first case then GFB using a nonmagical weapon does not bypass resistance to nonmagical attacks. If it means the second case, then it does. What really matters here more than all the posturing that both sides are engaging in is which meaning of delivered is being used.
I propose a simple test to determine which one is correct: evaluate whether a given meaning of delivered makes sense in the context of the two provided examples (assuming "another magical source" is supposed to function as a catch-all and not a specific example). The correct definition of the word will unerringly work with both given examples.
Case 1 - attack conveyed by:
Spell: The magic stone cantrip delivers magical bludgeoning damage through an enchanted rock striking the target
Magic Item: A +1 shortsword delivers magical piercing damage by striking the target
Case 2 - attack produced by:
Spell: GFB delivers magical weapon damage by granting attack
Magic Item: This is an interesting one. A Dancing Sword (DMG 161) grants its user an attack and would qualify as magical in this case. However, almost all other magic weapons (including the +1 shortsword used in case 1) do not produce attacks, and would therefore not qualify as magical if that is what delivered was intended to mean. When an interpretation means that magic weapons, on the whole, are not considered to deliver magical attacks I have no choice but to question the validity of that interpretation, especially when the alternative has no such issue.
I will let you draw your own conclusions, but for myself it is abundantly clear that the second meaning of 'delivered' is not the correct one, and as such GFB would not qualify as a source of magical bludgeoning/slashing/piercing damage. This follows the pre-errata phrasing, though that is not evidence for or against it. It is also possible that BOTH cases are correct, however I expect that "delivered by" would have been expanded to something along the lines of "produced or delivered by" if that was the intent.
Although my conclusion is in disagreement with many other posters in this thread, I hope that at the least it helps you all understand the reasoning behind my opinion.