- Saturday 31st August
For once, the first teletext e-mail request came through quickly and with
all 7 balls present and correct at 8.10pm. A shame then that my failure to log
out yesterday screwed up the generated graphics (and I doubt many people
would have looked at the [correct] ALT tags...). The update came with a
£9.6m jackpot estimate, which was revised at 10.34pm with an £8.5m
estimate. Amazingly, ITV teletext then dismally failed to update to an
exact jackpot amount throughout the entire weekend !
- Friday 30th August
Camelot were slow with their fax, which only arrived at 2.48pm today -
I duly updated the mid-week figures. A major screw-up on my part ensued as
I forgot to log out on Friday before I went home (the reason was that I had
to hurry to Vision Express to pick up some new glasses !)., so when the
Saturday evening screen dumps came along, they dumped the entire screen,
rather than just the balls - aarrgh !
- Thursday 29th August
Further RGB tweaks to lighten the InterLotto balls (UNIX Netscape tends to
dither the balls more darkly than their original values). I've probably
gone overboard with the green balls, which are now truly Day-glo :-)
Because of the InterLotto ball colour changes, I've now separated the
InterLotto and UK lottery navigation graphics directories because they no
longer share any of the graphics. Trimmed a little "fat" (i.e. redundant
UK lottery and virtual lottery code) from the InterLotto CGI binary.
I'm one sick puppy - for my Dad's birthday present today, I bought him a
lottery ticket containing 2 Lucky Dip sets for 8 weeks.
Added Steve Blundell's UK lottery page to list of sites from whom I
page grab on a Saturday evening - this makes 7 external WWW sites now.
Yes, I know of two more (one of which is the official site) which carry the
results, but they've proved elusive to parse to date.
- Wednesday 28th August
Changed the RGB values of the InterLotto ball colours to match their real-life
colours a little more closely (although the balls are "Day-glo", which is
tricky to get right with RGB colours). Announced the change in virtual lottery
payments (into an InterLotto account instead of by cheque) for draw #46
onwards in the newsflashes, but I'm holding back updating the virtual lottery
FAQ, rules and coding until the weekend.
- Tuesday 27th August
Uploaded the results for
Lottery #93. Fixed the ITV teletext page 123
parsing code so that it can pick up the jackpot pool amount anywhere between
the 1st and 10th word on the line, not that you can fit 10 words on a teletext
line :-) Fixed a minor bug in the
Number Palace that
wasn't including the latest two draws in the count of consecutive draws a
number had appeared in - this hit the number 44 this week of course, which has
just appeared in two consecutive draws a record total of 4 times.
Separated the UK lottery GIF cache from the InterLotto GIF cache (the latter
was just a soft-link to the former because I'd been using the same ball
graphics for both pages, which was incorrect really). Adjusted GIF cache
purger to purge both caches (UK = 16MB, InterLotto = 1MB) overnight on the
usual least recently accessed basis. Adjusted InterLotto GIF balls to be in
colour groups of 8 (i.e. 1-8, 9-16, etc.) rather than groups of 10 like the UK
- Saturday 24th August
The first teletext e-mail didn't have any numbers as usual, so the fallback
WWW grabbing code found Syndication UK's update at 8.05pm and Chris Prickett's
at 8.20pm, so it also updated the pages at 8.20pm. Note that
the wrong result for Lottery #92 (17th August) - they put "03" as the first
number, when it's actually "02" ! Annoyingly, Syndication UK put the
total prize pool on the same line as the estimated jackpot, so I didn't have
two (identical) estimated jackpots to work from and ignored them.
ITV teletext, in its infinite wisdom, decided to use a new format for the
jackpot prize pool line on page 123, namely "JACKPOT: £7,665,700".
This was the first time they'd put the jackpot pool amount as the second
word on that line, so my parsing code didn't like it when the first update
came through at 8.40pm. This meant the jackpot evaluated to zero and was
therefore removed from the home page for the rest of the weekend (even the
10.34pm or later updates didn't fix things).
- Friday 23rd August
Scanned in my winning cheque and put in on the usual "what I've spent"
- Thursday 22nd August
Received my £10 cheque for last Saturday's £10
win - expect a scan of it shortly. Did the usual
runs of page and software checking (spelling, weblint, internal links, gcc,
lint) and by studying the spelling checker output carefully, I picked up about
10 spelling mistakes, including some that had been hanging around for
months. :-( Camelot fax arrived at 9.23am, so I updated the mid-week figures
Adjusted the WWW page grabbing so that jackpot amounts grabbed that are less
than "100" are scaled up by a million. This now allows me to get Syndication
UK and Mersinet's fractional announcements of the jackpot.
- Tuesday 20th August
Installed lynx 2.5 on our systems and this allowed me to grab the
right-hand frame of Syndication UK
and parse their results on a Saturday evening for the first time (2.5 can
display some tables more sensibly than 2.4.X).
The September 1996 edition of .net magazine reviewed several
UK National Lottery WWW sites including:
- David West's pages (1 star - they
bitched about the wood background, which I actually think is the best feature
of the pages !).
official site (2 stars).
- Darren Presdee's pages (3 stars).
- Chris Prickett's site (4 stars).
- My site (5 stars, though they bitched about the background colours
Of course, .net magazine is the trashiest and worst-written of the
UK Internet mags, so I'm not sure getting a top rating is anything much to
shout about. Also, I've just noticed that Futurenet's WWW
pages (.net is part of that group)
require you to register...just so you can read last month's issue of
.net on-line (utterly pointless, but there you go).
- Monday 19th August
Uploaded the results for
Lottery #92. I
won my sixth £10 prize !
- Saturday 17th August
The first e-mail request came back at 8.04pm, but only had 5 of the 7 numbers.
Chris Prickett's site was updated by 8.15pm, but was the only one again
(of the 5 I check), so the usual return e-mail at 8.35pm produced the first
page update with a £9.4m estimated jackpot. The exact jackpot was
updated at 10.35pm.
- Friday 16th August
version of the 1.20 wheeling software and made it available.
Chased up Camelot about them not sending me a fax this week and duly
got one a day late. We've just installed a nicer error handling
system for "not found" URLs on the Connect server - here's the
page you get if you mis-type a lottery URL.
- Tuesday 13th August
Slightly improved the wheeling verification software (up to 1.20) in the
Wheeling Challenge section,
but it's nothing amazing really
(which is why I haven't hurried to re-compile it on a PC yet). The next
version will be 2.00 and will hopefully include code to verify a 4-match
wheel (which is probably a huge one !) as well. Updated some of
the info pages with the details from Camelot's second annual report.
- Monday 12th August
Uploaded the results for
Lottery #91. Sorted each category in the
Lottery Links section in descending rating order, which is
more obvious than the previous ordering I had (the order I'd put them in my
data file !). Added a new page detailing how to
The official lottery site didn't put the results on its
home page until
1.04pm today - a full 1 day and 17 hours after the draw...
- Saturday 10th August
The first teletext e-mail returned marginally too quickly (8.04pm) and only
had four of the 7 numbers, so was ignored. The first external WWW site to
have the results was, predictably, Chris Prickett's at about 8.15pm, but this
was the only one to update so was also ignored. The 8.20pm mail request
came back with a clone of the 8.00pm one, so I've shifted the second cron job
to 8.25pm (the same time as the first BBC teletext mail request).
BBC 2 teletext numbers came through at 8.36pm (along with ITV teletext's
estimated jackpot of £9.5m).
The first full set of ITV teletext winning numbers came through at 9.17pm,
but it wasn't until 10.43pm that the exact jackpot pool amount accompanied
a returned request.
- Friday 9th August
Received the latest Camelot Reports and Accounts (the second one they've
issued) - expect some updates to pages shortly, particularly in the
Background Info section.
- Thursday 8th August
Started to add links to the
Number Palace section from the individual lottery
pages, especially the "Draw Order" column and any references from now on
to particular numbers in the manual comments. Added a couple of lines of
draw order info to the Number Palace section, but it was tricky to phrase
them so it didn't sound like one number had appeared several times in a
single draw !
Received a new 168-ticket wheel for a
guaranteed 3-match for a 6 from 49 lottery, so I've plugged it into all the
usual places after it verified OK via the software I wrote a while ago.
Camelot's weekly fax arrived at 11.38am today and I updated the mid-week
stats as usual. They are also sending me info on any unclaimed jackpots
(only one at the moment) with the fax. Someone passed on a tip that helped
me remove the gap between table cell elements at long last !
- Tuesday 6th August
Improved the Number Palace section by adding facts'n'figures about individual
numbers - it fleshes it out a bit because the initial release was pretty
bare bones stuff. A change in the hostname of the Connect server a few months
back caught me out and I didn't adjust one of my scripts for the new name,
meaning that the dynamic GIF cache was trimmed excessively each evening
(128K instead of 16MB !) - it should be back it to its usual self now.
- Monday 5th August
Uploaded the results for
Lottery #90. Added a new
Number Palace section which has a fixed home
page and the rest is controlled with fairly routine CGI code. It doesn't
do anything spectacular at the moment really - I just thought that
some people might want to track particular pairs and triples, which
you can't currently do easily on other pages (the
Have You Won ? lets you do it, but
it's a bit clumsy - lots of to-ing and fro-ing required).
Had to cancel the virtual lottery again (the second time) - this
time there was a full disk last week that caused chaos with the entry log.
I'd put some error checking in w.r.t. not being able to open a temporary file,
but it probably opened OK and failed to actually write any contents. I've
now added many more write checks (including the return status of
every fprintf(), fclose() and stat() [updated file must be larger than the
original, which is an "obvious" thing I didn't check] on the log file)
in addition to the existing ones (failure to
open, failure to rename) - it's now probably as fortified as I can make
it...famous last words or what ? And, yes, I've now got a command that will
cancel a virtual lottery automatically...it's sad that I need one !
Swapped over the virtual lottery background colour (pastel green) with the
lottery home page background (pastel yellow) because I like the green
more and it's time for a change ! I also decided that the individual lottery
page backgrounds were too dark (aquamarine) and switched them to pastel blue.
Changed formula for estimated rollovers as follows:
- Find the "closest" similar pre-rollover draw [same number of rollovers]
via the smallest
difference between the amount rolled over from the current pre-rollover draw ("R1") and
the amount rolled over from previous pre-rollover draw "N" ("R2"). In other words,
abs(R1-R2) must be as small as possible.
- Find the jackpot prize pool (individual jackpot prize * number of jackpot
winners) for rollover draw N+1 and call that "J".
- Rollover estimate (rounded to nearest million) for the next jackpot prize
pool = (J*R1/R2).
I'm not saying this is the best algorithm by any means, but it's a little more
sophisticated than the last estimated single rollover (R1*2+1,000,000 rounded
down to the nearest million) and double rollover (R1+16,500,000 rounded down)
formulae. Oh, the normal jackpot estimate remains a fixed value (10m) in case
Removed the number selection strategy check from the returned
page from a Ticket Checker submission because
it would only make sense if one ticket was checked and you should be using
the more flexible Have You Won ?
page if you're doing that (which does always check your ticket against
my number selection strategy).
- Friday 2nd August
We had some serious problems NFS-mounting disks between today and Monday
morning. This heavily impacted all my automatic cron jobs (which rely on
NFS-mounted disks) and basically the page updates didn't work over the
weekend. For the technically minded, we think it's a problem between a
virtual interface patch to the kernel of the NFS fileserver and the
NFS-exporting of filestore.
- Thursday 1st August
Camelot seem to be sending me the mid-week results fax on Thursday
now (it arrived at 10.33am), rather than the Wednesday that they originally
managed to dispatch them.