@Zecc said:
Oh come on, Discourse is that bad.
Maybe if they figure out a way to scroll a page without creating gazillions of utterly useless history entries in the process.
@Zecc said:
Oh come on, Discourse is that bad.
Maybe if they figure out a way to scroll a page without creating gazillions of utterly useless history entries in the process.
@no laughing matter said:
Tried logging in and get "Unknown Error"!
o_O
10 PRINT "FACEPALM!" 20 PRINT "HEADDESK!" 30 GOTO 10
That must be at least near the top of the most WTFy things I've read here this whole year. My only reaction was, indeed, "WTF?"
Amtrak run rail services in $WHATEVER_COUNTRY_WE_AGREED_YOU_LIVE_IN?
The great thing about US rail services is they stop Britain from being at the bottom of the list when it comes to speed and frequency ;-) Plus the last bunch of locomotives we bought from you (or the US, depending on whether you're actually American or not) kept catching fire.
Unfortunately my interest in trains is also boring ;-)
I thought the Mr. Burns picture was due to a flaw in the Side Bar WTF panel of the website itself, i.e. Discourse is handing out the raw topic title, and the website itself is failing to escape it?
I have to wonder — if humanity survives a few more decades, where the IT industry will be, and whether we'll have finally made such bugs obsolete?
Wait a minute — what sort of alert are we talking about, that has no parameters of any kind?
It's interesting to see Aaron's suggestion at the beginning about dealing with open and save, since it did (for Apple users, anyway) come to pass.
@PJH said:
Hah!. If I could...
That's hilarious. It's also reassuring to know that I'm not alone in battling with a nightmare product.
@El_Heffe said:
@PJH said:
one (minor, admittedly) thing I like about Discourse that is missing on here is a like button.Please kill yourself, now.
I would have to agree with PJH. It's all too easy on a forum to write something that people appreciate, but they don't have anything meaningful to say in return. The result is that you think that you're being ignored, when you're not. Of course, if you really are unwanted and are better off dead, you can confirm this for yourself by affirming that nothing you say garners any likes ;-)
I'm glad someone pointed out that the new forum floods out your browser history with utter rubbish, as I'd not spotted that yet. I'm not going to have anything to do with Discourse as a result — not even read it. It's bad enough that Google Translate violates your history in much the same way.
Discourse also allows the same post to appear more than once on the same page: if you select to view replies to a post below a post, you'll just end up reading the same posts all over again as you scroll down. What sort of crazy mind does it take to conceive of something like that?
I don't have sufficient style to channel blakeyrat here — I'm just a boring nutjob (at least Ben L is entertaining). Seriously, if you've never used the old Apple System Profiler, you really won't understand.
It's a little bit like a Stockholm Syndrome of IT: it's all too easy after using Windows long enough to be mentally unable to conceive of anything better. You come to accept it on its own terms and that becomes the ceiling of your mental model. The only way to break out is to discover other operating systems (typically vintage ones, as a lot of ideas were born, grew and died in the 80s and early 90s), but the cost of this is that it shatters your peaceful idyll of the Wintel Wonderland. Linux is borderline here …
@RaceProUK said:
Never said they were broken. Although Dell isn't exactly a byword for quality.
What type of video card do you have? What bus does Windows say it's on?
@RaceProUK said:
@Daniel Beardsmore said:
If I understand you correctly, it is entirely Matrox's fault that Windows misreports AGP slots as "PCI"?More likely the motherboard drivers claiming the AGP slot is on the PCI bus.
In the two examples cited, one PC was installed using proper chipset drivers from Dell, and the other was installed in the Dell factory (although I've just upgraded to 8.1 this weekend). Apparently both have broken motherboard drivers that can't tell PCI, PCI Express and AGP apart from each other.
As with so many other things, if you tried Apple System Profiler from Mac OS 9 you'd understand. It's a level of enlightenment that cannot be taught, only acquired through experience. Sadly, the immense and irrational hatred tech folks felt towards classic Mac OS really hurt the industry, as people would shun all the wisdom on offer. I would never go back (one can grow tired of co-operative multitasking!), but what I gained from using it was worth all the pain and more.
@RaceProUK said:
The site you got those from seems to have a vendetta. Take this image:
We all know how this goes:
Windows: Hey Driver, what's that thing you drive called?
Driver: Hey Windows, it's called <name here>!
Windows: Thankyou!And then it's somehow MS's fault the driver gives the wrong name.
If I understand you correctly, it is entirely Matrox's fault that Windows misreports AGP slots as "PCI"?
You must be new to Device Manager. Let's see:
What's interesting to note is that Mac OS reported PCI device and vendor codes before Microsoft, who didn't get around to that until XP (Apple System Profiler featured quite detailed system information back in Mac OS 9 and earlier). Device Manager also has such delights as no USB device tree (Windows has one, but it only appears during certain device failures) and USB printers show up as "USB Printing Support" instead of the make and model of the printer.
You should spend some time around Device Manager. It's really quite enlightening .......
FWIW, that's an @font-face font, i.e. not one installed on the PC.
@The_Assimilator said:
OH MY GOD THESE SOFTWARE BUGS FROM 8 YEARS AGO ARE SO HILARIOUS HA HA HA ESPECIALLY SINCE THEY HAVEN'T BEEN FIXED IN ANY SUBSEQUENT VERSIONS OF THE SOFTWARE
Good. Use your aggressive feelings, boy. Let the hate flow through you.
I'm not sure what happened to that article date.
While it's altogether possible that some intern at Microsoft couldn't get their head around time synchronisation in Windows when setting up a new server (I can't be the only person who thinks that w32tm is immensely overwrought and broken), but that's a lot of clock drift even for a virtual server, and surely someone would notice long before it ended up that far out. I would also like to imagine that the server clock would be tied to that of the domain controller, and that the whole rack wouldn't all be that far out. (I don't know if domain controller time takes precedence even if the time difference exceeds w32tm's usual "too far out to be worth fixing, let's just live in the past/future forever" period.)
Yet, why would anyone be typing in the date manually?
Then you have MIME: the e-mail system takes the MUA's clock as authoritative. Outlook on the other hand uses the received date by default, so any mail that got delayed, shows an incorrect date and time.
As opposed to letting the GUI render the shortcut text based on its knowledge of what the shortcut actually is?
I really hope someone doesn't have to type those in by hand. Microsoft Minutes have taken a turn for the worse.
It reminds me of Crtl and Son of Crtl:
Yes, in Windows you had to (still do have to?) type menu shortcuts out by hand.
@Jaime said:
Because their first-level support is a bunch of idiots.
Their mail servers block any new IP address they've never seen before, as a matter of course. They then lie about a 24-hour response in their automated e-mail when you report it, as they seem to invariably take four days or so to do anything.
The best bit is when you ring them, and it's beyond the comprehension of the script-reading imbecile that it's possible to report a problem as (or on behalf of) the IP address owner, instead of a person with an Office 365 account who isn't receiving the message.
It wasn't the excellent support that I thought people said you got with Office 365.
I did in fact report the bug with the 0-second refresh of the 523 page yesterday. No reply and no sign of the website either.
@blakeyrat said:
It's also possible that you got a recycled bum IP from some other idiot and the whole thing's a false alarm. But spend the 10 seconds to check the router.
Is this the bit where I start ranting about how I never asked for any help or advice and that this is supposed to be a humo(u)r site, etc?
@El_Heffe said:
But the real question is what percentage of robots are for/against having sex with people.
The odds probably favour robots over computer nerds by a huge margin …
What, you mean my chances are as good as 17%? Things are looking up.
I can accept "every" as exaggeration for small figures, but as an exaggeration for "1", that's really pushing credibility a lot, whether you were trying to be funny or serious.
I am sure I'm not alone here in wondering why you set yourself up to get enraged at people. Maybe I should take it as a small comfort that I'm not the angriest person on the planet.
@blakeyrat said:
Yeah. Generally having every fucking internet site on the internet thinking you're doing a DDoS is probably a bad idea.
Every site on the Internet? Which other sites can you point to that share this concern? Looking through CloudFlare's case studies, I can access mitadmissions.org and www.eurovision.tv, which are both CloudFlare sites. In fact, I seem to have no issue accessing any of the websites listed there, except for one that is actually dead.
So, CloudFlare are only unhappy under the specific circumstance that very specific people (so far, two) want to see a specific site. A site that the affected people have never accessed before, and one that doesn't appear to be one that would be targeted.
FWIW I just ran a check on dnsbl.info — nothing there for my IP. If there is something on this PC, it would have to be a rootkit that hides both processes and sockets. Possible, but stupidly unlikely in this case.
I followed a link the other day to http://www.datawareinc.com/content/view/40/93/ and I got a CloudFlare CAPTCHA page come up instead:
Apparently if I run an anti-virus scan on my home PC, this will go away. How exactly would they know? What is their problem, exactly? I have not filled out the CAPTCHA — I doubt the page has what I wanted. (I was trying to get some more info on an unusual re-branded Nan Tan keyboard — I'm dreadfully sad like that.)
Today, I found a stranger one when going to https://encyclopediadramatica.es/Steve_Jobs:
Nothing particularly amiss there, except the page never stopped loading.
On closer inspection, they've set this:
<meta http-equiv="refresh" content="0">
The error page churns through "Ray ID"s like there's no tomorrow. I left it open in IE so that it would flood their logs with errors — it's been going for hours now. (It's more entertaining than discovering why anyone would use a photo of James Cromwell instead of Steve Jobs; I have no idea what that had to do with what I was actually looking for in the first place. Google image search was probably stoned.)
Interestingly, for that website, I don't need to enter a CAPTCHA, nor do I need to enter one to access the CloudFlare website. So what, do they think I was attacking Dataware?
@morbiuswilters said:
You lie, sir.
:(
I get the Flashblock "click for Flash" icon shown above Homoerotic Guy™. If I click it, a Flash ad loads. I don't block JS, and Adblock Plus only blocks very specific annoyances that mess up sites, largely site-dependent.
@joe.edwards said:
Awesome. Love the NSFW homoerotic ad banner. A++++ would click again.
I got that simply for using Flashblock, which I use purely for security. The real ad is Flash and should load above the dude with the cone.
Update, from Windows XP:
Opera 12.16: Aside from the terrible choice of font (like LCARS but square-edged), it works perfectly.
Firefox 29.0: Recursive frames, as per 28 in Windows 8.
MSIE 8: "Do you want to view only the webpage content that was delivered securely?" when accessing Google Translate; URL translation completely broken due to script errors.
From Windows 7:
MSIE 11: Frame error, as per 10 in Windows 8.
Make of that what you will.
@Seahen said:
Has anyone told Google that Translate could load Wikipedia over HTTPS, if they'd only create an account for it?
Would it help you if I made some nice pictures of all the websites that don't do HTTPS at all, that can't be translated?
@immibis said:
Is Google using user-agent sniffing to only set HSTS on Firefox? That would be very strange.
Well, consider that's also broken in Internet Explorer (screenshot above). Those are the only browsers I have installed (Firefox 28 and MSIE 10). I should have tried at work in Opera.
I remember the first time I encountered a breakdown in the British net nanny — some website used for hosting files too large to e-mail (I forget which) was falling apart due to an SSL certificate error, and this was only occurring on one network.
The cause turned out to be Thus's IWF (Internet Watch Foundation) proxy handing out its own SSL certificate to anyone accessing this one website, instead of the one for the website in question; naturally this makes browsers extremely anxious. As I understand it, someone had been caught sharing kiddy porn or some such over this service, so the government wanted it scrutinising for a while.
@Ben L. said:
No repro
You're using Chrome, so your address bar is still showing the site you tried to translate.
The bug is that Google Translate demands HTTPS or death. Going to http://translate.google.com/ redirects me to https://translate.google.com/ — if I remove the "s" in "https", Google puts it back again. Fortunately it doesn't redirect the actual translation URL.
This causes translation of virtually every single HTTP website to break (99.999% are not HTTPS). I'm finding this constantly now — every site I try to translate just gives me a blank page and that same useless error message.
My suspicion is that it really isn't an error message. It looks like Google are trying to politely warn you that they couldn't access the page by HTTPS, which seems a strange and pointless remark to make. I mean, you just accessed the website without HTTPS already — are Google trying to be some sort of retarded nanny? It's a page on NEC computers — does reading it make me a terrorist? Are Google trying to look after terrorists?
What I think happens in Firefox is that the fallback to HTTP fails; i.e. the blank page is a bug. Internet Explorer appears to be handling the situation better with an error page:
I suspect Firefox has a similar issue and, possibly, for once, Internet Explorer has the better error handling. Google simply don't care about testing in other browsers.
Here we go — "Load denied by X-Frame-Options: http://translate.google.com/translate?… does not permit cross-origin framing.":
The problem seems to be the meaning of X-Frame-Options: SAMEORIGIN — MSIE and Firefox (which MDN confirms hasn't got around to an error page yet!) disallow framing between HTTPS and HTTP with SAMEORIGIN set. This seems reasonable to me. I know Firefox and MSIE take mixed content very seriously now. I don't know if this is a bug in Chrome, that it's permitting this to occur? RFC 7034 does not actually say whether HTTP and HTTPS should be considered identical or different for the same host.
Of course, I don't actually know exactly what Chrome is doing here. I mean, Google are the people setting the X-Frame-Options header that's causing this to break in the first place. And what on earth are Google doing with HTTP and HTTPS here anyway? I am not clear what they're trying to do exactly that's wound them up in this mess in the first place.
Ah, I get it. Google forces me to use HTTPS (via redirecting HTTP to HTTPS), then won't translate any sites that don't use HTTPS!
However, if you remove the 's' from "https", the translation comes up just fine, without another redirect back to the HTTPS site.
Basically Google's left and right hands aren't talking — the error message now seems like a sulky retort to whoever it is who's forcing HTTPS when they know full well that this just breaks pretty much every website, hence the lack of blamestorming in the message, since Google Translate is wilfully causing its own problem.
For Japanese to English itself, Google and Bing are pretty much equally awful. Japanese seems to be monstrously difficult to translate.
Not sure what's gone wrong this time …
Trying to translate a Japanese Wikipedia page that contains a lot more content than its English-language counterpart:
Google think that it's terribly helpful to tell me that “This page was not retrieved from its original location over a secure connection.” What is that supposed to mean? Does the “Dismiss” button dismiss whoever was responsible for yet another useless message?
There's a secure connection that refused to hand over the page? I'm banned from translating anything that's not on a secure connection? What? What am I supposed to do with that message? Whose fault is the problem? It's almost as bad as the diabolically dreadful “I could’t give a toss, so go FOAD” passive voice “Document not saved” from Excel, that sends me into murderous rage.
Something seems to be messed up with Wikipedia too. I figured maybe if I tried HTTPS it would comfort whatever gremlin in Google is deciding I'm not allowed to have my page. Instead I got this:
Recursive framesets, yay!
Depending what you click you can get Google really confused:
I'm still no closer to getting a translation of that page.
I think it may have been a lexical file handle.
If you want to be picky, the error should be: "Can't open settings file $somefile: $!" — all too often, I know that some file somewhere can't be read, but I have no idea where it is …
Imagine a configuration file like so:
setting_a => 42 setting_b => 10.100.4.5 setting_c => 1338 setting_d => Z
I don't have the code to hand, but it looked a lot like the following (as best as I can remember):
Of each " => ", the " " and "> " are removed at the top, and the "=" is removed later on — not action at a distance, but … just … why? The chap helpfully factored out a bunch of string replacements, but didn't think to factor out splitting the line into key and value. In reality, the if statements did slightly more (also not factored) and were denser and harder to follow (LOTS OF STUFF IN CAPITALS FOR NO REASON), but I forget the specifics now.
Also, he's found Perl's foreach
, yet he's using it solely to terminate what is otherwise a while loop.
I don't know what Perl version this was written for, and it may not have had switch statements (given
/when
) but some elsif
s would have at least restored faith that the guy had some idea what he was doing. Then again, the only reason I was called to look at this code was after a crash caused by a miserable lack of error handling ('Cannot call method "login" on an undefined value' after a connection timeout).
This codebase is always good for a laugh. It's not good for a lot else …
Bonus points:
@skotl said:
AAARGH!!!! Not that one!!
The developer responsible for that deserves a whole new level of Hell specially creating for them.
@dhromed said:
@DaveK said:
Dhromed-purple-dildo-joke incoming in 3....2....1....I've grown since those times.
I R DISAPPOINT
@anonymous234 said:
Remember this thing? I really wish it had gone somewhere:
Yes, you can emulate it with a tablet or touchscreen, but it's not the same as actual keys.
Those are Cherry ML switches. You may consider this a blessing or a curse; a lot of people complain about off-angle binding with those (the wider key spacing won't help here), and they're pretty scratchy, (although I'm one of the few people who quite likes ML switches). Some lube should sort that out …
The Optimus Maximus did go into production, by the way.
@jmfridey said:
Hi All,
It took me a while to cut this one down. I created a blog post on my website with detailed instructions and links to the updates you'll need to download. It'll take you about 10 minutes and you should have Windows Updates working again.
http://www.geekallday.com/windows-xp-sp3-update-issues/
Splendid, dear chap. That did the trick.
(In theory — I just ran the missing steps on this here butchered XP computer, the one where I'd run some bizarre update agent installing contraption, since you're the first person I've seen who's found an actual download link to the agent software. Windows Update is now working perfectly, and very fast at that.)
@PJH said:
@Daniel Beardsmore said:I have only one word for whoever designed this Heath Robinson contraption: "gloves".These ones?
Aye.
Apple Events are the message-passing IPC (inter-process communication) system for Mac OS/OS X. It's how programs talk to each other, locally (and someone must have used this feature at some point) over a network, and how the operating system delivers high-level messages such as instructions to open or print documents, or quit. It's like a binary version of D-Bus, as unlike RPC/RMI, you're not calling real functions (although allegedly, with OS X and Cocoa you can now do that); rather, programs simply inspect the message's ID codes to determine what to do.
For example, when you select a bunch of files in the Finder and double-click one of them, the Finder creates an 'odoc' Apple Event message, and inserts a complete list of files into the message. The receiving program is handed all the file details at once, so you don't get the Windows vicious mutex dance. An extra event type of 'gURL' was introduced at some stage to pass URLs to a program. Suite codes are provided for namespace purposes.
If you have a text editor that has no idea what a paragraph is, the operating system's IPC mechanism is the least of your worries.
Obviously I'm not going to go into extensive detail of Apple Events/AppleScript/OSA/OSAX etc in a forum topic. Essentially I'm just making you and others aware that ideas and alternatives exist, and did exist, that may not be present on the operating systems that you're familiar with.
It should also be fairly apparent to you that none of Apple's decisions, past, present or future, are binding on any other system; you're not expected to bow down to the shrine of Jobs, but rather, learn from both the strengths and the weaknesses of other systems, past and present. For example, D-Bus is seemingly a completely independent invention along similar lines, that takes a different approach to solving the problem of soft-coded message-passing IPC.
BTW, Apple Event functions and methods use named parameters, which back in the 90s at least, had to be specified in the correct order. You could omit non-applicable parameters, but not specify them in the wrong order. That restriction aside, they're a lot like command line parameters.
@dkf said:
That's what the manual is for; remembering all the little details so you don't have to. To find things in the manual, use
apropos
or look online (of course).
That's the whole point. In a GUI, the computer tells you what you can do, and gives you an obvious means to do it.
With the CLI, you have to remember the name of the command, then bring up the argument list, scrutinise it for five minutes trying to remember what exactly it was you were supposed to type in to get the exact effect you want (as programs rarely "do the right thing" by default, or they do a lot of things and you need to specify exactly what you want — think of the obstinate replicators on Star Trek that demand exact parameters first). Then you have to double-check whether long args use - or --, make sure you've checked whether the key verb parameter takes a dash or not (e.g. apt-get update does not -- it's not apt-get --update or apt-get-update) — all sorts of strange syntax contortions to trip over.
Best example: Robocopy. Instead of "doing the right thing" (as Explorer's copy/move operations mostly do these days), you have an argument list of Tolstoy proportions, with all the sensible options off by default, and absurdities like infinite retry for fatal errors such as permissions problems. (Explorer's approach is getting quite good now, but it's still got serious problems of its own, especially when dealing with MAX_PATH.)
wget is another one where you have to scrutinise the argument list every time you want to do anything besides download one small file — how do I ensure retry is set correctly, how do I ensure it doesn't create a load of pointless subdirectories for the domain and URL path, things like that. While a GUI tends to keep the parameters contextual (or offer a wizard), a single command-line utility can have a massive argument vocabulary that's so time-consuming to analyse that the end-result is far more time is wasted than if you'd used a graphical tool.
Command lines are fantastic for the most common tasks, but for anything that you want to do infrequently, the burden of understanding and specifying the parameters becomes excessive.
Apple had an interesting idea, with Commando in A/UX — it would present all the parameters to a UNIX command in a dialog box:
(From Toasty Tech — Apple A/UX)
@anonymous234 said:
Editing the BCD is a major pain in the ass, so you best hope it doesn't become corrupt. If any part of the process before that is broken, a "quick and dirty" solution that sometimes works is just grabbing any Linux disk and letting it install GRUB.
I don't know whether I mentioned this here at the time. A week after purchasing a Windows 8 PC, which I hibernate each night, the hibernation process broke the BCD. I would get a sad face BSOD immediately on boot.
One problem was that the BCD file was not where anyone said it was, not with UEFI anyway — it was trivially accessible within Knoppix if you knew where to look, but to get Windows 8 to see it (to run bcedit to remove the broken hibernation entry), I needed to mount the hidden partition with it on. I also had no Windows 8 media to begin with, required to get the recovery console up. The end result was that it took me a week to resurrect the computer. I learnt that my old P4 laptop might suck for the Web, but it runs Inkscape quite competently (and the 1400×1050 screen also helps).
I have only one word for whoever designed this Heath Robinson contraption: "gloves".