What's New On These Pages: October 2003

Next month: November 2003

• Saturday 25th October
  Camelot introduced three new machines named after (semi-)precious stones - Amethyst, Opal and the boringly-like-a-lump-of-rock-named Moonstone. I had fun looking at stone sites for much more exotic names such as Sapphire, Aquamarine (reminds me of Stingray) and Topaz, but I must say I like the dry humour behind the name Sardonyx the best of all :-) Oh, the site code has been adjusted to cope with the 8th, 9th and 10th names (with 3 more to come in a month or two according to Camelot) used for main/Extra draw machines.

• Monday 13th October
  It's nice to be able to log in at around 10.00pm on a weekday evening and see your site nicely up to date with the latest Daily Play numbers without even having to lift a finger :-) Anyway, a quick look at the logging for tonight's Daily Play updates showed that the earliest cron job at 8.35pm actually picked up the numbers ! No surprise then, that I'm starting the cron jobs at 8.30pm from tomorrow onwards...

• Saturday 11th October
  I've added the final tweak to the Daily Play auto-grabbing script, namely that as soon as it comes across a valid machine and ball set for an evening's draw, it will "remember" them for the rest of the evening, even if they "disappear" for a while (yes, this does actually happen on the official site regularly). This nicely prevents up to 2 unnecessary sets of page generations, which since they take a couple of minutes each, is a useful bonus. Of course, this assumes that their first appearance is the correct one and they don't change later on...

  I've now stopped generating the playslip row and column analysis pages across all games. Even though I'd actually removed the links to the pages weeks ago, I was still creating them after every draw. It should knock a few seconds off of the overall page generation time, which is always a good thing. Oh, and I phoned Camelot again this evening for the full results because the official Web site was crazily slow again (45 minutes to put up most of the draw numbers - except for Daily Play - and 75 minutes overall to release the full results).

• Wednesday 8th October
  I don't know if Camelot think it's acceptable to take an hour to put up the winning numbers for any of the draws tonight and a total of 2 hours to put up the full results of said draws, but I certainly don't. I phoned Camelot to see what was going on and they seemed to have the full results on their phone line before the official Web site, so it looked like a Web-only issue.

  I phoned up Camelot this morning to ask them why Linux users are still barred from using three of their regular links on the official site's home page (to be precise, it's the "Sign In", "Open an Account" and "Subscriptions" links). It turns out they have a "spreadsheet" (huh? Not a database?) of users who can't use those links (I guess it's mostly Linux users) and if enough people get on their list, they'll consider supporting the affected platform(s). I'd rather they put up a warning like "you are running an unsupported OS - we cannot help you if you run into problems using this area of the site" and let you continue, than just block off significant parts of their site to all users who don't happen to run OS'es manufactured by a convicted monopolist or by a vendor of overpriced hardware/software (I spent £220 setting up a brand new (pre-built) PC this year with Linux on it...can the fruity company get anywhere near that price...didn't think so).

  I was well confused by the new "online subscriptions form" because it gave the impression that you could susbcribe to the main lottery completely online. It turns out that it's actually just using a combination of your online account details (yep, I don't have those cos I use Linux and can't register !) and the numbers/subs period you chose to dynamically generate a PDF which, wait for it, you print out and post to them ! It was hardly worth the effort coding this really because they've had an "unfilled" subscription PDF for download for ages now (which, ironically, Linux users can no longer access on their site, whereas they used to be able to before this "semi-online" effort was put on the official site).

  Got some word (no names or sites will be mentioned) that Camelot are finally chasing up potential "lookalike" sites (probably mainly those that use official logos) and telling them to take down any copyrighted material and state on the home page somehwere that they're not an official lottery site.

  Finally, for those on tenterhooks dying to know how the Daily Play updates did, well they worked fine for getting the winning numbers, but the 2 hour delay got me frustrated and I hand-updated the data file after phoning up Camelot and getting the number of winners in each tier from them. A quick run of the Daily Play page generation script by hand and Robert's your brother's father as the saying goes.

• Tuesday 7th October
  Although the Daily Play updates went smoothly, I had some calls to a midweek_draw() function when deciding whether to put the 8-seconds auto refresh code on the home page. Of course, this should only be on a Wednesday, but some midweek draws have actually been on a Tuesday in the past, so the function returned true for tonight and duly made the home page auto-refresh every 8 seconds :-(

  For the home page stuff, I have made the code only check for Wednesday (and Saturday in fact of course), ignoring Tuesday, which should fix this problem for future Tuesday nights. Apologies to anyone, as P. Fiennes pointed to out to me via e-mail, who had their "dry-eye condition or migraine" set off. Yes, I don't like pages that never bring up any new info when they regularly auto-refresh - see most of the pages on Sporting Life (even static match reports that never change have a 5 minute refresh on them !).

  Oh, as for tonight's Daily Play auto-updates, I actually managed four tonight would you believe it ! There were just the winning numbers at 8.41pm, the machine/ball set added at 8.48pm, then - most bizarrely - the machine and ball set were actually removed from the official site at 8.51pm, followed by the full results (including the machine and ball set this time) at 8.54pm. Unfortunately, my code doesn't "remember" the machine and ball set if they appear and then get removed again later. Seems that the official site seems to be quite inconsistent as to how they put up these Daily Play results.

  Bought two more tickets for tomorrow's rollover main draw.

• Monday 6th October
  Added a lock check to the Daily Play auto-grab script so that if any page generation (for any of the games I keep a data file for) is in progress, the cron job backs off until all page generation locks are clear. This won't make a difference until Wednesday's draws of course.

  What was pleasing about tonight's 3 (yes, three) separate official site Daily Play updates (winning numbers, then machine/ball set, then the prizes/winners) is that the auto-grab code caught all three and updated the site correctly...and I didn't even check the pages until 9.15pm or so (20 mins after the third update). So, hopefully, I can leave it even longer tomorrow night before checking in (and if it works on Tuesday and Wednesday OK, I'll probably do no more manual checks that it's worked on non-main draw days).

• Saturday 4th October
  No problems with the Daily Play update tonight, though the issue of two sets of page generation running simultaneously (the cron one for Daily Play and the manual ones for the other draws that I do) on Saturdays and Wednesdays is a problem (because Camelot are tending to update the Daily Play info at around the same time as the other draws). The solution will be to check for page generation locks for the other games in the Daily Play cron job and back off the cron jobs if they are in progress.

• Friday 3rd October
  Strangely, the official site's Daily Play updates changed their order again tonight - it was the winning numbers on the drawn order page, then the machine/ball set on that same page and finally the individual draw page was updated with the full results. Anyway, I've now completed code to cope with any order of updating after the drawn order page gets the winning numbers, so it should now run fairly smoothly hopefully.

  Like a bull in a china shop, a German user (IP address 217.83.158.175) decided to run a spider (which disguised itself as IE 6 on XP, when it probably wasn't) at 4.56pm UK time yesterday. By the time it had finished some 82 minutes later, it had made 12,060 requests (not a single one for an image of course), which was a rate of 2.45 requests per second.

  6,486 (53.8%) of those requests ran lottery CGIs, which infuriated me because I have long had a robots.txt file to specifically exclude /cgi-bin/ for all spiders (to prevent heavy machine load and any issues about "infinite spidering"). It annoyed me enough to actually post a newsflash up about it on the home page, though I don't think I can be bothered reporting it to Deutsche Telekom this time. And, yes, it set an all-time hourly request record for this site ! If the German spider-fan is reading this - I won't stand for this happening a second time...

• Thursday 2nd October
  I studied the official site's updates more closely than usual tonight - interestingly, they put the full results up on the appropriate Daily Play individual page (and these included the machine and ball set), but it took several minutes before the "last 6 months in drawn order" Daily Play page included the machine and ball set. Because I don't look at the full results page unless the drawn order includes the machine and ball set, the auto-update was stalling.

  To fix this, it looks like I'll have to get the full results page as soon as the drawn order page gets its winning numbers, so that I can parse in the machine and ball set from the full results when they appear.

• Wednesday 1st October
  The Daily Play auto-grab code worked fine for the winning numbers (minus any machine or ballset), but then seemed to refuse to pick up the full results, which I ended up entering into the data file by hand and running the re-genration script manually too. I shall investigate this problem of course.

Previous month: September 2003

© Richard K. Lloyd & Connect Internet Solutions Limited  2024