The Official Status Thread
-
Yeah which is retarded because it works fine in C:\Program Files. Which is where I put it. Like, it sets itself up to install in the wrong folder, but since it fixes permissions on-install so it can auto-update, there's literally no reason for them to install in a non-standard location.
-
I put it on E:, where I have most of my other (few) games, because my C: is a bit smallish.
Status: Updating game client: receiving updates (23%)
-
I'm showing it's about 13 GB installed, BTW.
-
Does that include any DLC? The installer says it's downloading about 5GB.
-
Well the shit it downloads is compressed of course, and it keeps old installers around. So as you patch its folder gets larger.
-
World of Warships is a free download.
Thanks, but I needed the nap.
I will keep it in mind for another day.
-
Status: Bad shit happened at my company yesterday. And since my boss promised to assign tickets to me and hasn't (my JIRA queue is entirely empty) I think I'll just "work from home" on my Xbox today.
After working from home from for like two months now, I can tell you, that's often how it goes.
Hope they unclog that toilet soon.
-
Status: whenever people say Marina Sirtis on Star Trek TNG was a bad actress... well, she was kind of, but she did a great job in "The Survivors" when she was suffering from the psychic "block" song in her head.
-
Status: 502 OK — from just looking at the monitor, without even touching the keyboard or mouse.
-
And... dialed into my last meeting of the day, but it's been 10 minutes now and still just hold music, so I think it either got cancelled or someone forgot to open the conference line.
Popped off an email to the meeting organizer and I guess my commitments are done for the day. Whee.
-
TIL Oregon school shooter has a page on MySpace, of all places. Seems appropriate.
Hey, this exposure could be their big chance for a comeback!
-
Reality: It just took that long for the post you tried making a few minutes ago to come back as a failure.
-
Got something like that too.
And a few hours before, a message that my post couldn't be published for reasons, but it appeared nevertheless.
Status: telling my internalized mommy that I'm not tired.
-
During conference call, crazy new age QA guy is trying to do a guided meditation before we get down to business, this has been going on for 9 minutes now
That's what farting into the receiver is for.
Even if they figure out who that was, claim it was clearing your toxins and flowing your chakra or something.
-
Status: Why the fuck does one of my developers do fantastic at things I have time to review and completely botch everything I'm too swamped to review until it's gone out to UAT? Seriously. Typos, misspellings, etc. Not just in the code, but in the fucking APIs. And that's before the obligatory race conditions.
-
Status: Going to bed before 8:00 PM. Can't wait until my jaw heals up enough that I can start eating normally and actually having energy again. (Not to worry anyone, I just pulled several short straws in a row as far as bad luck with wisdom teeth extraction and I will likely not be back to full strength until early next year...)
-
Seriously. Typos, misspellings, etc. Not just in the code, but in the fucking APIs. And that's before the obligatory race conditions.
I can forgive everything but sloppiness. It's like the fuckers just don't care at all, have no pride in their work. They just want to bang into keyboard until shit works and go home.
I hate it.
-
STATUS: Remember how, like years ago, TDWTF was making fun of go-s package management? How it could only pull master off github and had no way of shuffling multiple versions of the same package inside the same project?
Well, seems like, as of a few months ago, you can finally activate an experimental switch and get some kind of sane dependency system going. Oh, except the native tools still don't recognize it, and you need to go digging through like 78 third party patchwork little tools, to help you set up your own crappy ad-hoc system for managing this shit.
Coming from NPM, it's like stepping off my golden chariot and landing my shiny polished boot into a pile of manure, while dirty toothless villagers snicker from sidelines.
-
dirty toothless villagers snicker from sidelines.
Don't make fun of Ben's dwarves like that...
-
What, this?
I finally watched it. That is Captain Planet. You can call him something else, but that is Captain Planet. The only thing he is missing is a self-righteous diatribe about recycling.
-
Saving that clip for when we have a "customer just sent unprecedentedly large shitheap of a data file and choked out the memory on the production boxes" incident. We usually have one or two of those a year.
-
If it worked but was a mess it'd be fine (or at least less bad) but fucking typos in hard-coded integration critical strings?
-
It's Captain Planet, isn't it?
-
you just got an error when trying to submit if they weren't filled in.
I have, in the past, dealt with APIs that would respond to errors by just giving you the site's home page instead. Because. Thankfully, not this time. This time, it's trying to do automated uploads of a file (filled with metadata about the bulk archiving I've just done ;)) but the there's no way to do that except by pretending this monster of a form has been submitted. The developer of the service says that an API would be a good idea, but is too overworked to sort it out. “Pull requests accepted” has been uttered, but it's a complicated knot of Rails so like fuck am I going to try figuring out why the automated API generation doesn't work for POST.
Also, the XML version of APIs generated by Rails is shit. I'll rant about that some other time…
-
- Me: "Hmm, I don't fully understand the meaning of the Linux setuid and seteuid system calls, I wonder if a quick internet search will clarify my thoughts?"
- Google: "Here is a 20 page document explaining in great detail how those system calls are too confusing for you to ever understand"
And I still can't find any clear confirmation of whether the ONLY way to elevate a process in Linux is the setuid permission bit.
-
And I still can't find any clear confirmation of whether the ONLY way to elevate a process in Linux is the setuid permission bit.
The only way? no. but it's the only way i know that doesn't involve abusing security holes to elevate.
As i recall the Setuid system call is used to downgrade permissions, and the seteuid call is used to set your effective permissions to a lower state, letting you revert to elevate status later.... i think. it's been a while since i did systems programming for linux so i might be misremembering or possibly the rules changed.
-
It just seems really hackish even for Linux standards.
...or maybe it shows that my standards for Linux were too high.
-
well the other way to do it usually is to not use setuid bits and just make the user elevate to root via sudo or su before invoking your program.
also linux? not being hackish?
BWA HA HA HA HA HA HA HA!
-
well the other way to do it usually is to not use setuid bits and just make the user elevate to root via sudo or su before invoking your program.
That's unrelated. Invoking via sudo or su replaces the setuid bit in the filesystem for your program.
The setuid system call is to drop privileges once the special resource (raw device handle, low-numbered TCP socket, etc) has been acquired. Preferably at that point you change both the real and the effective UID to your unprivileged ID, so you can't regain root permissions.
-
That's unrelated. Invoking via sudo or su replaces the setuid bit in the filesystem for your program.
that's not what i said? huh...... i thought i was clear there.
-
Ehm, yeah, I think I misread a couple posts. The bit and the system call having the same name is confusing.
-
The bit and the system call having the same name is confusing.
tell me about it....
there are three things that are hard in computer engineering:
- Naming things
- cache invalidation
- naming things.
-
The problem is now you have to start a new process, and do your elevated stuff there, and in some circumstances that can be hard.
For example, let's say you're making a high level scripting language, and you want to abstract the whole elevating thing by adding a special keyword that runs a particular function or piece of code as root (after getting user permission). You'd have to launch a new interpreter via sudo, transfer the necessary program state there, and continue the execution there.
I suppose the official Linux stance on that is YOU'RE DOING IT WRONG.
-
Google: "Here is a 20 page document explaining in great detail how those system calls are too confusing for you to ever understand"
The moment I see "Abstract" at the start of a document:
-
You can't post that as a png
-
Huh? What happened? I was pasting something from imgur, when I fell asleep.
-
For example, let's say you're making a high level scripting language, and you want to abstract the whole elevating thing by adding a special keyword that runs a particular function or piece of code as root (after getting user permission). You'd have to launch a new interpreter via sudo, transfer the necessary program state there, and continue the execution there.
That's very difficult to get right anyway. In fact, it's going to be just about at the top end of things that are difficult to get right in general unless you're making the [spoiler]goatse.cx[/spoiler] of programming languages. It'd be easy enough to do if you were dealing with a pure functional call that you wanted elevated, but there's no reason to bother there. Once you're outside that limited domain, what happens to the state is going to be a… challenge to all sorts of things. It makes my head hurt to think about the edge cases and so on; I think I won't bother.
No, what you've used as a use-case is something that belongs on the Bad Ideas Thread.
-
It's a good idea, it's just current OSes have no sane way of supporting it. Even ones without the stupidity of the "setuid" bit or whatever accalia was jabbering on about.
Don't confuse "OSes don't support it" with "bad idea", that's how you completely kill innovation in software.
-
Status: Twitter is broken in a new and innovative way this morning. They seem to have accidentally installed a stylesheet which prevents the "tweet" button from appearing.
-
Well, that is something that probably should have been caught by automated testing...
-
Well, from a security perspective you do want to be as certain as possible what parts of the state are going in and coming out, to minimize attack surface.
-
Status: Salesforce recently released an OSX version of their dataloader tool. Figured it was worth a go over the third-party tool (which is no longer available now anyway). Downloaded it and went to run it - it tells me I need to install the Apple legacy Java ONE POINT SIX JRE package to run it, despite Java 1.8 being installed.
-
Reminds me of the advertisement I received recently:
Dear teacher,
you'll be happy to know that we updated our lesson material which previously only covered MS Office 2007. It's now updated for Office 2010!Our school is using 2013 and we're in the process of evaluating how to get our staff and pupils free licenses and thus 2016.
-
STATUS: Finished some bitch work around my flat.
There. Now just to commit the changes...
...
WTF brain!?
Filed under: It would be nice, though
-
@PWolff said:
and wondering why getting used to bad practices is beginner friendly.
It would be bad practice in some languages. But if having everything default is a feature of VB, this isn't bad practice there.
I mean, in LISP, it's good practice to use dashes in variable names.
Sometimes different languages really are different.
I happened to think about this again. After thinking about it, I agree that it can be "good practice" to rely on initialization of variables to well-defined default values on declaration.
But the VB guys didn't do that consistently:
Dim objVar As Object testSub(objVar) 'Gives a warning about a possible NullPointerExceptionFor i = 0 To 5 - 1 Dim j As Integer Debug.Print(j) j += 1 Nextprints the numbers [spoiler]0, 1, 2, 3, 4 in spite of the fact that j is only visible within the loop and that one should assume it was reinitialized at the definition point[/spoiler]
-
Status: Been kicking ass at World of Warships all afternoon-- keep getting paired with shitty teams that don't.
Killed a destroyer, 2 cruisers and a carrier in a destroyer. Despite punching WAY above my weight class, that team (natch) lost and I got shit for XP. Ugh.
Except with the Japanese Myogi. That thing's a piece of shit. Only 3 turrets (two in the back, meaning if you're not broadside, you ain't shit), 30s reload time, and accuracy so low you can't hit shit.
-
Don't confuse "OSes don't support it" with "bad idea"
I was actually thinking about what information was being passed between security domains.
Basically, the process boundary is the principal security boundary in modern operating systems — they're pretty much defined by this — and so any time you do anything that's really robust security-wise, you've got to use a process boundary in there somewhere. (There's also VM boundaries, but they're much more heavyweight.)You can do it also by some sort of internal sandboxing, but that's hard to get right, especially in the subordinate-elevated case. Identifying that the subordinate code may run with elevated privileges, do just what it is permitted to do (what does that even mean?) and nothing else is where it gets brain-bending. Proving correctness of general code (which is what this comes down to) is astonishingly hard.
Perhaps it is not impossible. But the chance that someone will fuck it up is colossal. Splitting the problem into several pieces makes it quite a bit simpler. But not simple. I know because I've written code that does this. It's difficult and subtle. And if anyone knows how to do something like this on Windows without dialog boxes popping up, I'd be interested in finding out more. The solution I've got for Unix definitely doesn't port, and
runasdoesn't seem to guarantee no-blocking-dialogs so far as I can see.
-
it tells me I need to install the Apple legacy Java ONE POINT SIX JRE package to run it
There are two possibilities. Possibility one, the code's a pile of shit that was written against 1.6 but doesn't realise that that wasn't the latest version because some idiot used
!=instead of<… Possibility two, they're doing one of the few things that really requires 1.6 instead of anything later; there's a few subtleties here and there in places like SSL handling and bytecode generation which might cause problems.Please let it be possibility 1. Then you'd just have ordinary incompetence. The other option is the higher-order incompetence of someone who knows how to do neat stuff, but not why you should stick to simple things as much as you can.
-
Open a support case with MS. That honestly sounds like a thing that you shouldn't be doing but I cannot articulate why. If we're lucky, Raymond Chen will tell us what obvious thing you missed.
Wait. Is your problem not solved by impersonation? Or do you need to do shit that isn't impersonation aware?
-
Status: Slept in this morning, then two errands before coming home for a nap. Maybe I'm more sleep deprived then I thought I was.
