This particular patch provides a subset of the original arguments. I can live with that theoretically, but it also swaps the positions of pk_items and updates. It took some time to find why my .update(id, {"a": "b"}) was broken 🙈 And the fact than my IDE (PyCharm) shows me hints about the original method and "go to definition" also jumps to the old one, doesn't help with such cases.
Hi!
This particular patch provides a subset of the original arguments. I can live with that theoretically, but it also swaps the positions of
pk_itemsandupdates. It took some time to find why my.update(id, {"a": "b"})was broken 🙈 And the fact than my IDE (PyCharm) shows me hints about the original method and "go to definition" also jumps to the old one, doesn't help with such cases.I understand that that "@patch"ing stuff was made for the sake of nbdev and it works in live environments. But such quirks with incompatibly changed method signatures make using of FastHTML in a Starlett's typical "hot reload" workflow pretty uncomfortable 😩
Maybe You can consider to make the args "keyword-only"? This will make the patching more robust.