When putting together the new server I uncovered a severe inefficiency in the code of that page. I won't bore you with the details (although I personally find it enthralling) — but let's just say after using SQL for 30+ years I only just learned a very important nuance about indexing timestamps.

So the new server had the new upgraded super-fast code for the Recent Kibitzing page. I didn't see anything amiss, because when only me and a few associates testing the server there really was no kibitzing to see.

However, in the excitement of the efficiency upgrade, I forgot how the logic of the page works. Eons ago it really did wipe itself clean at midnight, but at some point we changed "today" into "everything since midnight yesterday" so that there was always something there to read.

So now I've updated the page to have the efficiency improvements *and* retain the logic of the "since midnight yesterday" clause.

I had assumed that you just used a time-slice window for all post younger than 24 hours (or whatever).

E.g. an even more general cut for posts older than a day but younger than two days:

<
SELECT *
  FROM YOUR_TABLE t
     WHERE
      t.datetime_co- lumn < DATE_SUB(NOW(), INTERVAL 24 HOUR)
     AND
      t.datetime_co- lumn > DATE_SUB(NOW(), INTERVAL 48 HOUR)
>

http://stackoverflow.com/questions/...

Even my Android phone can do those... (that is, if I had an Android phone, or any phone for that matter).>

Here's the big revelation: the query used to read something like this

SELECT * FROM Kibitz WHERE TO_DATE(timestamp) = '2016-10-12';

It looks perfectly reasonable. TO_DATE() simply takes a full blown timestamp and reduces it just to the year/month/day portion. But the problem is that TO_DATE() obliterates any indexing on the timestamp field. So enormously faster is this awkward looking construction:

SELECT * FROM Kibitz
WHERE timestamp >= '2016-10-12 00:00:00'
AND timestamp <= '2016-10-12 23:59:59';

Get it? "Anywhere from midnight Oct 12 to 11:59:59 pm Oct 12." A longwinded way of saying "Oct 12" — but 1000 times faster! And of course that can be extended for a two-day or longer window as well.

But if we clarify the context and how it works, and hopefully a few more people request the feature, I'm open to it.

Even if there wasn't a single malicious hacker on earth, you STILL should use placeholders and bind-variables, just so the database can cache your queries!

Thanks for all your efforts to upgrade hardware etc.

I think I noticed one slight omission now where it indicates on a game:

"Featured in the Following Game Collection..."

https://web.archive.org/web/2016040...

It indicates at the end :

"plus 547 more collections (not shown)"

But this is not on the new page for that game.

Could that bit be reinstated please?!

<SallyS> posted this link in a post:

Botvinnik vs Euwe, 1948 (kibitz #26)

http: //www.chessgames.com /perl/kibitzing?kid=G1032171&reply=26

which gets translated to this:

Botvinnik vs Euwe, 1948

http: //www.chessgames.com/perl/chessgame?gid=1032171&kp- age=2#reply26

(verbatim mode would be handy here - i.e. respect all linebreaks and whitespace, and don't auto-magic links or fens, etc.)

<The problem with the link translation is that I don't see the post on the screen - instead I see the game score, ending at the Tip line.> That's a problem as old as the hashmark was used for page anchors—which is to say, since the WWW began. The problem is that your screen cannot scroll so far down the page to show Sally's post, because the page just isn't that long.

If it were, then I wouldn't be able to hand scroll down to the post, right?

Please take another look and correct me if I'm wrong... but the link is to post #26, and the page goes down to kibitz #50 (for me).

I think the problem is more due to the kibitz of interest being first in the page.

<Jul-22-05 aw1988: Upon the insistance of the Soviet authorities to gain a win over Euwe at any cost, ...>

Yup, that's sounds more likely.

As you suspected, I don't see the problem with Firefox, but it consistently shows up with Chrome.

(I'm typically using Chrome on Windows most days recently, but often run Firefox as well...

I have Opera and Safari on older machines, and will check at least the Opera later today.

http://www.chessgames.com/perl/ches...

Indeed there was a problem searching for names less than 4 letters. Do to a cock-up in the configuration file the directive to allow for 2-letter words was being ignored, and it lapsed back to the default of 4 letters.

While I was at it, I noticed the very ancient code that stripped out 2-letter words from name searches, so that "Dr. Lasker" would just be "Lasker" etc. That code dates back from when we were unable to search on 2 letter words, so I disabled it.