- RdSplay code de-uglification, part 1. Fixed one of the three XHTML errors I know about and made it possible to embed a feed item list anywhere (another page, sidebar, even this post if I wanted to).
- pArse update for Atom titles.
- Minor addition to RdPress: a last comment link, added to the home sidebar.
- RdMise Auth logout code: it's not pretty, but basically it keeps sending 401 responses until invalid or no login info is sent.
- Made paths try matching case-insensitively. If that succeeds, the browser gets sent a redirect.
Still not perfect, needs debugging.Works.
Still to do
- Fix at least one more bit of invalid XHTML (the menu on pages with child nodes that are hidden causes an empty <ul>).
- Make stuff work better with the new auth code.
- Last-minute porting over of the anti-spam code.
Cycle 4, Day 16
My 8th chemotherapy treatment was yesterday. Went fine, I suppose, except that fact my left arm didn't like it and I had to get another cannula put in the right one. Oh, and the nurse missed the vein first time. And I... uh... misplaced my lunch...
Oh well, I suppose the fact that I feel well enough to use the computer this time is good. I've got a CT scan booked for Day 3 next cycle (that's January 13th to the rest of you) which should show if I've been wasting my time.
- The spammers haven't got anything through for ages, but they're still trying.
- I can now successfully get to the login page on β with my phone. Then I try to go to any other page, and I'm not logged in. Suggestions welcome. (Going back to cookies doesn't count, though I realize it might be my only choice...)
- I have a WordPress blog running as a test, with "
Disallow: /" in robots.txt. Now Technorati is picking up my posts from there as well as here, and Feedster doesn't seem to be showing posts from the real site at all. I'm guessing it's partly my fault for not noticing that WordPress pings Ping-o-matic by default, but still, I thought robots.txt was pretty clear.
Sometimes, I hate being right...
[27/12/2005 15:16] <tH> hmm
[27/12/2005 15:16] <tH> yesterday, the spammers started limiting themselves to one link per post
[27/12/2005 15:17] <tH> but i already had their domain blocked so none got thorough :)
[27/12/2005 15:17] <DreamEater> so they are watching your site then
[27/12/2005 15:17] <tH> no spam for 24 hours now, so either they've given up or they're going to try to catch me off guard in a few day's time
Well, it wasn't a few days, but sure enough, 68 comments just got through whilst I was eating. Spam deleted, blocks improved, and the cycle starts again. Yawn.
Fixes and workarounds
A few changes on β:
- Made the top level navigation show across the whole site.
- Gave in and added the live bookmark icon to the feed links. Decided to keep them orange. Made the main link huge, adding another rule to the IE-only stylesheet in the process.
- Replaced the →/⇐/⇒ entities with the numeric versions, which is a good idea in general and a workaround for a Thunderbird 1.0 bug. I also made this change here after being reminded by Hannibal.
- Worked around a bug in my phone's browser's support for HTTP Auth. (It wouldn't accept
algorithm=MD5, only the incorrect
algorithm="MD5". Solution: realize
algorithmdefaults to MD5 anyway and remove it.)
- Fixed a bug in my parsing of the
Authorizationheader. Augmented BNF makes my head hurt.
Login working on my phone is the last thing I want to fix before putting the new code on the real site, and I'll check that again once I actually have some credit. :/
1 to ChatZilla
And whilst I'm posting, I might as well note that bug 321195, which randomly makes some ChatZilla motifs ugly in Firefox 1.5, was filed and fixed. No idea which Firefox release it'll show up in though, maybe 18.104.22.168, or 2.0 if we're unlucky.
Just got a huge spam attack. At least a thousand comments over a day, mostly drugs (including one I actually have on prescription :/), posted via what looks like a botnet.
I have to admit, that's a relatively clever thing for a spammer to do: set off an attack near Christmas, whilst most other people are busy spending time with friends. Infer what you like from that.
*Goes to patch his parser*
Current state is ✘✘✘✔✔✔✘✘.
Edit: ✔✔✔✔✔✔✔✔, though the only way I survive XHTML is by ignoring it, and I still fail on Phil's actual feed, since my faux-HTML parser understands < but not <.
A Bunch of Fives
In no particular order, I suppose.
TV Shows I'm currently addicted to
- The Daily Show With Jon Stewart (More4) - Ironically, probably the most reliable US news source available here.
- QI (BBC TWO/BBC FOUR) - Everything you ever didn't need to know, from Custard to Cuttlefish.
- The Late Edition (BBC FOUR) - "It's 'Newsnight' with jokes."
- Space Cadets (Channel 4/E4) - Like Big Brother in space. Except, not really. In space, that is.
- Lost (Channel 4/E4) - I've watched every episode since this started, and it still manages to fuck with my head each week.
Podcasts that rock
- Digital Detroit Radio (XML) - Some great independent music, with funny stuff inbetween.
- The Ricky Gervais Show (XML) - I was suspicious of this at first, but it turns out it's just Ricky Gervais and a couple of friends talking almost as much crap as we do on Numbcast. Warped sense of humour required.
- My Virtual Band (XML) - Collaborative songmaking. Sounds crazy, sounds crazy.
- Off The Wall/Hook (XML) - Tech/politics stuff you never knew you needed to know.
- Extra Super Action Show (XML) - "I should kick you right in the neck." :|
I've been wasting time too much recently.
Pointless fact: 89% of this site's legitimate referrers are search engine results pages, and 87% of those searches are on Google (3% Yahoo, 8% MSN). When I'm next in bored stats mode, I'll compare the amount of visits I get from crawlers to the amount of traffic they give me, and then scan the query list for anything interesting.
I also need to update the Clear Bug View script and decide if I want to start using a User-Agent blacklist on this site.
Update: Now working for Opera and Firefox on β. Just need to attempt IE support, then add it to the real site.
Update: Works. Still ugly in IE, but I'm not bothered for now.
Update: In use. (XULRunner page)
RdMise 1.1 (odd=development) is up at β.rdmsoft.com. The code has not changed very much in the past 3 months, and it's running off the same database, however there are several reasons I'm doing this now:
- I'm playing with an IDN to get the β in there. However:
- IDNs don't work in IE.
- Since that whole spoofing thing, at least Mozilla and Konqueror don't trust IDNs in .com (I checked Opera 9tp1, and .com isn't on the whitelist, but my IDN seems to work anyway :/). If you want, you can create a boolean pref in Firefox called "network.IDN.whitelist.com" and set it to true, or mess with the KDE_USE_IDN variable, but it's probably not a good idea to trust the whole of .com just so I can have a pretty URL.
- RdMise doesn't have a clue about IDNs, so it uses the xn--nxa address whenever it needs the hostname. I doubt this will actually be a problem, though I haven't read any specs yet.
- The new template/style (menu in top-right, but after main content in the code) needs testing, particularly on my phone. I also don't know if the login will work.
- The TV guide is disabled (and has no database), but I want to get it running on this server soon.
- All requests to that subdomain (except /robots.txt and /favicon.ico) go through the script. I decided against this when I first put up rdmsoft.com, opting for /r/ instead. I want to see what problems, if any, it causes.
I know there are still some links pointing to /dlfs/ and /imgs/ that won't work on β, but please let me know if you see any other problems.
Okay, on to the real reason I'm posting this. I was the 24th person in Beverley and Holderness to sign up, and the system needs 25 before it contacts Graham Stuart. Could somebody in the area sort that out?
Now I really should add support to my parser for type=html and type=xhtml on <title>s, even though I doubt the latter will be used much. It's already messing up one feed's titles in my reader, and is probably more annoying than the odd encoding issue I'm still having.
See, that's what a good spec is about - punishing developers for their own laziness, and making sure they can't blame anyone else.
- Update my CZ/XR patch. The zip download's already updated to 0.9.69, but I need to fix a few prefs in the patch and make it zip automatically.
- Add UI for the ChatZilla source search.
- Rewrite my IdentD plugin so it has per-network settings. I don't think plugins can have per-network prefs, so I'll probably use an array and save to either a single pref or a text file.
- Add "starring" to my TV guide app with some DHTMLAJAXBBQ™, then upload a test copy of RdMise to see if I've broken anything yet.
- Stop procrastinating and code a proper solution to invalid XML in blog posts. Also, maybe change the font so I don't keep typing <Ii> instead of <li>.
- Get PHP to compile and APD to work.
KPF is dead, long live Sunbelt KPF.
Whilst I've never used any of Sunbelt's products, their blog makes a good impression and they seem pretty competent. Besides, they must be doing good work to get so many legal threats from the spyware companies...
Exactly a year ago, I posted some site statistics. I kept analyzing them for a bit, then got bored, but I figured I might as well have a look at last month's stats and see how things have changed in a year.
AWStats for last month came up with around 35% unknown for OS and browser stats, so I've decided to ignore those hits for these purposes.
- 87.5% Windows XP and 2003
- 5.5% Older Windows
- 3.4% Linux
- 2.3% (Free|Open)BSD
- 1.2% Mac OS X
- 0.1% OS/2
Windows XP numbers are up, at least compared to older Windows. Both Linux and FreeBSD came higher than OS X, which doesn't really surprise me. (Hello, OS/2 users.)
Minor browsers ignored, inaccurate, skewed, yadda yadda...
- 39.9% Firefox 1.5
- 26.8% Firefox 1.0
- 6.5% Mozilla Suite
- 0.2% Netscape 7/8
- 20.8% Internet Explorer 6/7
- 2.5% Internet Explorer 5
- 1.6% Netscape 4
- 1.2% Opera
- 0.4% Safari
- 0.1% Konqueror
Firefox 1.5 being the most popular browser seems strange (It was only released on the 30th) until you take into account that the three release candidates were (correctly) labelled as 1.5. I also had a few hits from 1.4.1 (1.5 beta) and 1.6a1 (trunk nightly build) which are basically the same, so I counted those as 1.5.
Of the Firefox 1.0 visitors, 85% were using the latest security update (1.0.7). It'll be interesting to see how long it takes before Firefox 1.0.x becomes ignorable, from a web design point of view.
Nothing surprising here, except maybe Netscpe 4 beating Opera, and Mozilla Suite being higher than usual. I don't know if AWStats recognizes SeaMonkey yet, I'm presuming either no visitors used it, or it got counted as Mozilla Suite.
Referrer spam is up. A certain feed directory which I won't name is my top non-human source of referrer headers, requesting my feed 200 times whilst advertising its own URL. Malice or incompetance, I'm not really bothered.
About 7% of visitors get here via a search engine, and last month about 1 in 10 of those got here by searching for [firefox 1.0.8], which I haven't written about, and doesn't exist yet. Still, that doesn't stop me being the first result on Google. :|
I haven't bothered looking through the full list, but I'm happy in the knowledge that my website is helping a few people per month uninstall Windows Media Player and find out how crap Sony music players are (Sidenote: What's the plural of Walkman? Walkmans?). Somebody also got here because they wanted to find out what the "useless flag" does in BZFlag. Y'know, I thought that was obvious...
About 80 hits per day, which is bigger than before. How many are humans and how many people actually read the content, I obviously can't tell.
That's about it for now. I recently noticed I was in the top 10 on Google for my own name, which is a more or less difficult thing to achieve depending on your name. I know "Robert" is very common, and "Marshall" isn't rare, but it's at least better than, say, "Smith". :P
Oh, and third for [crap snow]. How depressing/appropriate.
ChatZilla 0.9.69 should be out soon, the first real new release since March. It includes one patch by me (to fix the -chat command-line flag for Firefox 1.5).
ReadMe fixed shortcut keys, sp3k added an infinite connection retries mode, and Hannibal and Silver did far too much to mention. You'll see it all when it's ready, which should be this Sunday, December 4th.
My source viewer now shows up-to-date trunk, and should be refreshing daily thanks to the magic of cron, cvs, and php (previously I was generating the highlighting and search index at home, and uploading it all to the server, which was less than optimal). Trunk is currently marked as 0.9.67+, but as far as I understand, it's very close to what will be released as 0.9.69.
ChatZilla-related items on my todo list at the moment: fix the source viewer CSS, rewrite the IdentD, update XULRunner patch, and then maybe get round to actually working on ChatZilla itself...