WTF Bites
-
662 users had password «123456», 168 — «123», 115 — «password».
1409 users (1%) used their username as their password, in its original form, without any modifications.
10% of users reused their leaked passwords: 9.7% — directly, and 0.6% — with very minor modifications.
I don't have enough s, especially given how many popular packages were affected.
-
@raceprouk said in WTF Bites:
662 users had password «123456»
Stupid, you need to go up to 8 to be secure !
-
One of the users directly controlling more than 20 million downloads/month chose to improve their previously revoked leaked password by adding a ! to it at the end.
That should work too, as
123456!
is a much bigger number.
-
@timebandit said in WTF Bites:
@anotherusername said in WTF Bites:
And there's no way for an installer to say "hey, I'm done but I require a proper reboot sometime when it's convenient", which forces it to do a real shut down instead?
No need, it just wait for Windows Update to do its thing
Windows Update: ERROR: Cannot apply changes with a reboot pending. Odds? (I've seen some software that will do that)
-
Why would I need to know any of that?
-
@hungrier If Justin is too old or wearing a toque, you don't want to tell him you love him in any language
-
Why would I need to know any of that?
First of all I would have to ask her what a toque even is XD
-
First of all I would have to ask her what a toque even is
I had to look it up. It's a bobble hat
-
First of all I would have to ask her what a toque even is
I had to look it up. It's a bobble hat
Ah. Then I don't need to ask cortana about it because I'm definitely not ever going to need one
-
@akko nobody needs a bobble hat, but many people's lives are enriched by them
-
They're useful in Canada, but this is the wrong time of year for them.
-
@akko nobody needs a bobble hat, but many people's lives are enriched by them
Let me rephrase: "Ah. Then I don't need to ask cortana about it because I'm definitely not ever going to want one"
-
I guess Microsoft thinks I'm very old. Opened the calculator for the first time after reinstall and this was the size it opened at:
Thank you Microsoft for being considerate, but I don't really need the extra large buttons. Especially when I use the numpad to control it!
-
@atazhaia I don't see the problem. Your computer is a 5" phone right?
-
@atazhaia You need large button if you use the One True Way to control things - touch.
-
@hungrier I suppose Windows could get my phone and my computer confused. Both have a 1080p screen and 6-core CPU after all, as well as mainly run a Linux-based OS.
-
Improper email validation strikes again!
Additional notes:
- This is a trial install, not purchased
- You can't skip registration
-
@dreikin and the installer embeds a webpage...
-
If you configure a service to run as
<username that starts with a number>
, systemd will launch that service as root instead. Because in their opinion usernames cannot start with digits, and because realistically usernames containing only digits cannot be distinguished fromprocessuser IDs.
-
Windows 10 calculator
Maybe there's no room there because every device is a phone for fat-fingered giants, and that's a pretty small window with no wasted vertical space whatsoever. I'll just resize it and surely then the button will fit.
What if I make it wider?
Oh hey, there's a whole other tab that isn't in the smaller version. That's pretty good design right?
What if I go even wider?
The buttons changed places. That's pretty good design right? I guess that first layout was more optimized for a taller orientation and this is more square, so maybe it changes again to accommodate a really wide window.
Edit: I closed and reopened it, and now it's got a title bar:
-
@hungrier the modern calculator has always had responsive design, and as long as you have the hamburger menu you can access all features regardless of screen size. TR is the missing title bar.
-
version
Oh hey, I didn't notice Windows must have wasted bandwidth pulling an update for an app I don't use!
Also...
Hey! I want my Aero effect on the entire app! Where can I get this???
-
TR is the missing title bar.
No, TR is using the calculator in Standard view.
-
@hungrier the modern calculator has always had responsive design, and as long as you have the hamburger menu you can access all features regardless of screen size. TR is the missing title bar.
You're right about the title bar, and I still don't know why it happened in the first place. But as far as responsive design, changing to a vertically oriented 4x6 with super wide buttons instead of keeping the 5x5 square or going to a horizontal 6x4 is anti-responsive. Also, the memory tab isn't in the hamburger menu, so that part is locked away if you keep the window small.
-
-
@wharrgarbl Who in san francisco wants more rats?
-
@yamikuronue said in WTF Bites:
@wharrgarbl Who in san francisco wants more rats?
Pet rats are very different creatures to sewer rats:
In case you're waffling about having brunch with rodents, the Humane Society of the United States says that pet rats are "highly intelligent and social animals who clean themselves regularly and thrive on regular interaction."
I have a friend who used to have pet rats, and they really were the sweetest little fuzzballs
The rats, not the friend
-
title bar
I've found out how to trigger the disappearing title bar: Lock the computer with the calculator open and let the screen turn off.
-
Lately Windows has been popping up these low memory messages despite having plenty of memory available:
-
Lately Windows has been popping up these low memory messages despite having plenty of memory available:
Ah, Windows 95 called, wants its resources back!
-
Lately Windows has been popping up these low memory messages despite having plenty of memory available:
Do you have virtual memory size capped? It seems to me that you have 98.7% comitted, even if the physical isn't full.
-
@swayde I do have it capped, but based on the actual usage it isn't even hitting the limit of my physical memory, let alone virtual. Are there some under-the-hood shenanigans going on where it's actually using all that committed memory or what?
-
@hungrier I'd guess that was the problem. Win 10 does a lot of magic behind the scenes. As you see in the screenshot it compresses memory, and propably tries to page it out and fails.
The first thing i'd do was to let windows manage the size, and reboot.
I have no idea how theese numbers are meant to sum
-
reboot
That would help anyways. Windows is probably hiding a slow memory leak.
-
-
I have no idea how theese numbers are meant to sum
Even funnier is the "recommended" setting
Anyway I've increased it from that, hopefully after I reboot all my problems will go away forever
-
@hungrier
for 24GB psysical ram:
I've never touched that setting on this install.
-
@hungrier What they really wanted to say was "Close Opera and use Edge instead"
-
Q: When is a null-check not a null-check?
A: When it's in C#! ...?
(source)
-
-
@arantor
It's not like your users can really enter NULL anyway, an empty string should be enough for anyone.
-
@djls45
Seems sane. It's essentially a performance optimization, and is abstracted away - this is something the .net CLR does. If this is important to you, you are meddling with dragons, and you should know that.
-
@swayde My issue is not necessarily with basically ignoring a null until it causes an access violation; I wonder just how it can backtrack to just before the null-check and take the alternate path, especially if there are "side effects" like I/O that occurred between the null-check and the
NullReferenceException
.
-
@djls45
I presume they only do this for side effect free code, or chunks smaller than what can be kept in L1/L2. AFAIK intels newer (1-2 gens) CPU's has some kind of hardware optimization for exactly that case. (not just pipelining). I'll be damned if i can google it up however. They had to disable the feature on the cpu it was originally launched with because it was buggy.
I must admit it's not something i'd like to implement, or test for that matter.EDIT: i was thinking of TSX, which is not related (but still interesting)
-
@swayde My issue is not necessarily with basically ignoring a null until it causes an access violation; I wonder just how it can backtrack to just before the null-check and take the alternate path, especially if there are "side effects" like I/O that occurred between the null-check and the
NullReferenceException
.And wouldn't that make the case of an occurrence of a
NullReferenceException
hugely more expensive than a simple null pointer comparison?
-
@djls45
The obvious answer would have to be no. Since they are doing it, it must be performant.
It might be a case of do the work we have, or waste the time instead.
Not null:
We execute the instructions we have, presuming the object is not null. This saves a branch, and the associated loads from main memory (if we're really unlucky).
Null:
We execute the instructions we can, after we've executed the instructions we discover that we've wasted 10 cycles doing something useless, but we're going to have to wait for a load/store/decode from main memory anyway. We've wasted 10 cycles+ 90 to load from main memory. And we have to spend time unraveling the stack.Depending on how theese factors relate to each other, and how often the not null assumption is true (~90% would make it even in this example), we might gain performance.
If we have to raise an actual exception, my guess would be that it's several orders of magnitude more expensive than doing the naive calculation.
Note that i don't actually know where i would find this information, but i'd like to know more in depth.
It could be that they check the pointer at the absolute latest point, like described in this article:
@dkf might know, if it's a smart trick, the .net guys can't be the only ones doing it.
gcc seems to have "-fdelete-null-pointer-checks"
-
@swayde Good analysis. I suspect that in my code 99% of null checks are "can't happen" cases and/or there to satisfy a static code analyzer. That's just anecdotal but I assume the .NET guys reached a similar conclusion given their choice of implementation.
-
@djls45
The obvious answer would have to be no. Since they are doing it, it must be performant.
It might be a case of do the work we have, or waste the time instead.
Not null:
We execute the instructions we have, presuming the object is not null. This saves a branch, and the associated loads from main memory (if we're really unlucky).
Null:
We execute the instructions we can, after we've executed the instructions we discover that we've wasted 10 cycles doing something useless, but we're going to have to wait for a load/store/decode from main memory anyway. We've wasted 10 cycles+ 90 to load from main memory. And we have to spend time unraveling the stack.Depending on how theese factors relate to each other, and how often the not null assumption is true (~90% would make it even in this example), we might gain performance.
If we have to raise an actual exception, my guess would be that it's several orders of magnitude more expensive than doing the naive calculation.
Note that i don't actually know where i would find this information, but i'd like to know more in depth.
It could be that they check the pointer at the absolute latest point, like described in this article:
@dkf might know, if it's a smart trick, the .net guys can't be the only ones doing it.
gcc seems to have "-fdelete-null-pointer-checks"
Since C# is managed, couldn't it be the case that the VM has special-case code the compiler can use to say "on null dereference, do this instead" instead of actually having to do an exception?
-
Status: Whoops, seems I forgot to update this VM into SP1...
-
@heterodox said in WTF Bites:
@swayde Good analysis. I suspect that in my code 99% of null checks are "can't happen" cases and/or there to satisfy a static code analyzer.
A static analyzer that insists on null checks all over the place for impossible cases where the runtime would just throw a perfectly fine exception otherwise is Broken. Because adding code like
if (arg == null) throw new ArgumentException("arg")
does not really improve quality of anything, just takes time. And if those null checks are impossible, you can't do anything else than throwing another exception anyway, can you?