U Atakisi vs E Berg, 2007 (kibitz #15)

You folks who own/maintain this site <do> in fact pick many from 2007.

Well, then, would he <please> suggest Kool Aid and Oreos for lunch??

Chumbawamba!

https://www.youtube.com/watch?v=2H5...

Oh Danny boy..

<That's what I mean by "ill-defined." Whether PGN is valid isn't a binary proposition. Some are unquestionably invalid; some are impeccably perfect; but most fall into a spectrum where they quality is acceptable, but not a good as it could be. Exactly where to draw the "acceptable" line is a matter of debate.>

OK, I'm trying to avoid working on another topic - so this is a good diversion.

I really don't care in an absolute sense what threshold you use for the import PGN in terms of correctness.

You picked a good example with:

<1.d4 d5 3.c4 c6 2.Nf3 dxc4>

I don't have NICBase, but I do have SCID and CBLight, and both will blithely accept that input.

But the essential fact of the matter is that they then <regularize> the PGN for export.

In both cases the PGN is corrected to read:

<1.d4 d5 2.c4 c6 3.Nf3 dxc4>

And in truth, my concern is that <CG> itself doesn't promulgate abominable PGN - i.e. that the only PGN which is downloadable from <CG> is *legal* PGN, and super-squeaky clean PGN at that.

It's really not such an unreasonable request now is it?

As far as the rest of the post - you're indeed talking about the difficulty of what you're willing to accept, and correct.

That's between you and your doctor/priest/lawyer/rabbi.

I have opinions on it, of course, like always. Given a game submitted by a generic user - any irregularity should be bounced, and the user asked to submit an improved version.

As far as adopted the pre-existing games from a large DB to "seed" <CG>, that's a different matter - since there's no user to explicitly ask.

Right - we have two cases. One it's fairly easy to decide to be stricter and more likely therefore to be correct.

So, bounce the submission <1.d4 d5 3.c4 c6 2.Nf3 dxc4> back, with the expectation that <1.d4 d5 2.c4 c6 3.Nf3 dxc4> will be returned. Put the onus of "getting it right" on the submitter - where it belongs.

<Chessgames itself is a PGN validator.>

Well, sometimes it is, but sometimes it isn't - that's my experience.

I've found lots of invalid PGN, as you well know.

<During this stage we have enforcements that are not part of the PGN standards—just for example, we do not allow for "?" in the Date field, although it is 100% legal to do so. Our standards are higher than typical commercial software because we're not just checking for validity, we want to make sure it has a minimum quality.>

Of course <[Date "?"]> is legal, and proper PGN. What else does one do when the date is unknown?

OK, maybe you might insist on a standard format like this:

<[Date "????.??.??"]>

but the idea is the same. If you don't know, you don't know.

* * * * *

Once again - I want to emphasis - there's two items at issue.

One - what PGN will <CG> accept?

That seems to be the focus of your response.

Two - what PGN will <CG> export?

<1) +0.12 (34 ply) <2...e6 2.d4 d5> 3.Nc3 Nf6 3.Bg5 Be7 4.e5 Nfd7 4.Bxe7 Qxe7 5.Qd2 O-O 5.f4 c5 6.Nf3 a6 6.g3 Nc6 7.dxc5 Qxc5 7.O-O-O b5 8.Kb1 b4 8.Ne2 a5 9.Ned4 Nxd4 9.Qxd4 h6 10.Be2 Ba6 10.Qxc5 Nxc5 11.Bxa6 Rxa6>

Similarly, after 1.e4 e6 2.d4:

The generic linking isn't as informative as I'd like.

Of course you shouldn't give the move away - most people will be posting the position as tacticals would be my wager.

You could do a <(after <move>)>, but why not keep it simple and parrot the url?

I don't like it; it's non standard. As far as I know, 20.5 does not mean "Black's 20th move" anywhere else in the world except in Chessgames software and URL parsing.

Maybe if it was just

<D Howell vs R Jumabayev, 2016 (20...)>

But if we're going to do that we might as well just say 20....?

There's a pretty fast server (32Gb RAM, quad core) which is processing opening positions with Stockfish around the clock. I started off doing 90 minute computations, then hour long thinks, then switched to 30 minutes, then 15 minutes, then 10.

Well, since you invented the notation, I figured when in Rome...

It's not a bad notation, btw. And using it would help familiarize people with the notation - since they have to handcraft the url at present.

But, being more verbose is OK. That's good too, <White's nth> or <Black's nth>. There's always the confusion of whether it's the last move or the move to be played, of course.

So then the explicit, <(move n, White|Black to move)>, solves that.

Over on Bang I've adopted the convention:

(White|Black to move after ...)

And list the "pre-move" (i.e. the last move, like 1.e4 or 1...e4).

Thanks, morf