username change: do older PGNs change?

Hi,

A corner case question I imagine, but in the case of someone changing their username and  chess.com continuing to supply player_id in the Profile endpoint, will archived games fetched with the new username (the only way to access them) contain the old or new username?

The question is not an idle one: if the old username appears it won't be recognised as a game by the player whose games are being sought, therefore the PGNs should change.

A definite argument for retrieving games when wanted and not caching them "too long" for whatever "too long" is.

Giles (who changed his username at signup when he made a typo, and before playing a game )

Usernames do change.

Interpreting that to mean that if I download PGNs post-username change, the "White" or "Black" tag will have the new name.

Thanks. It really couldn't work otherwise I suppose.

Yeah, I've been downloading PGNs for over a year without any issue, even for people who have changed names.  Now, I haven't been using the API, but I would imagine it gets the same PGNs.

The only difference I've seen is the addition of a "Link" tag to give a URL to the game on chess.com. Absolutely excellent any time a particular game needs to be referred to and there are multiple between the two players on the same day; common in blitz.

Oh, and the web interface appears to give \r\n terminated lines in PGNs for the blank line separation of tags from moves and moves from the next game. I'm not sure that that isn't a bug rather than a feature: either no bare \n or only bare \n.

I've coded to \r?\n so don't have to worry which I have or even if it's consistent.

<alert>topic drift </alert>

The Link tag would occasionally be useful, but I've always managed without so far.  \r\n doesn't matter to me.

Granted I haven't needed the Link tag either, but I've wanted it even so.

\r\n is small potatoes these days; each side of the war remains convinced it is correct and both handle either format.

Now, Unicode (specifically, as UTF-8) in PGNs is against the specification, but in this situation the specification should be ignored until it is updated. It should also get a bug fix and regrettably for backward compatibility and because people presumably occasionally write one by hand can't be simplified to ease parsing.

I can confirm that the data is calculated using the actual username, so if a player changes his/her username all data will reflect the change.

Thanks @andreamorandini. It really couldn't work any other way without exposing a history of usernames which isn't done, but nice to get the confirmation. Thank you.