WTF Bites
-
That is indeed where Teams belongs.
Luckily, all we use it for is some video conferencing. Chatting is via Slack. Some meetings are still Zoom. And it has behaved itself on Ubuntu so far...
-
@DogsB what the fuck is a situationship?
Why do these kids keep making up words?Zoomers and millennials are idiots and need a new word for everything because they don't know what a dictionary is.
Yeah, because whatever boomers remember from their childhoods is eternal tradition. Everything else is just a temporary fad, unlike their temporary fads that are just how the world works, or would have been if not for those meddling kids.
We have a description for this situation. Mutual attraction.
I think it was called something else 50 years ago. Crime perhaps? Or maybe just an affair. I don't know, internet doesn't have any good guide on boomer teenage slang. By Gen X years, I'm pretty sure it was called booty call, but might need a correction on that.
Yeah, what’s that word for not reading the article?
"Geoblocking"? At least I think that's what they're doing when they won't let me read anything but the headline.
-
I borked the git submodules again, in a different way. I removed (with reset --hard origin/master) a commit from a submodule, not realizing that it was already pointed to by a commit in the parent repo. I then added another commit, updated the parent repo, and pushed everything to origin. And nobody noticed that there is a bad hash somewhere in history, the repo could be cloned and the submodules updated.
That was until the gitlab runner failed withfatal: remote error: upload-pack: not our refFortunately I still had the bad commit on my local copy and merged it back (and reverted) with the original SHA. I have no idea if this could be fixed in another way without losing history. Would deiniting, manually deleting and reiniting the submodule work?
-
@sebastian-galczynski I think it wouldn't. You'd have to rewrite the history of the parent repo too.
-
@sebastian-galczynski I think it wouldn't. You'd have to rewrite the history of the parent repo too.
I'm not sure. In the parent repo we have deleted submodules which no longer have an upstream, and this doesn't cause problems. So wouldn't deinit and init create the same situation?
Edit: And why exactly is upload-pack causing problems, while high-level git fetch runs fine?
-
@sebastian-galczynski Deinit & init wouldn't affect the history anyway, so if it got stuck on the revision referencing the bad commit, the module was still initialized there, so it would still be stuck.
The modules that no longer have upstreams don't cause problems because you are not pushing new commits referencing them. I think it would recover from this one too as the later revisions have valid references, possibly after restarting something.
-
Talking out of my ass here, but wouldn't just creating a new branch or tag referencing that bad commit be enough to ensure it doesn't get gc'ed from the submodule repo?
-
Talking out of my ass here, but wouldn't just creating a new branch or tag referencing that bad commit be enough to ensure it doesn't get gc'ed from the submodule repo?
Just pushing the branch, without merging it? Yea, probably.
-
The modules that no longer have upstreams don't cause problems because you are not pushing new commits referencing them. I think it would recover from this one too as the later revisions have valid references, possibly after restarting something.
The problem is that it didn't. I only pushed after adding a commit with valid submodule hash, but with an ancestor with invalid hash. And that broke the thing gitlab was doing (but not pulling by other devs). And those no longer existing submodules are also in its ancestors somewhere, and yet those don't cause errors.
-
@sebastian-galczynski said in WTF Bites:
I removed (with reset --hard origin/master) a commit from a submodule
Since you deliberately ventured into the clearly marked
Danger Zone , your troubles aren't too surprising. Rewriting published-and-depended-upon history is a recipe for trouble, and should only be done when you've got at least a chance to track down all the places where you need to clean up. Which using submodules can make more difficult... 
-
@sebastian-galczynski said in WTF Bites:
And that broke the thing gitlab was doing
Which thing exactly? It probably isn't doing that thing for the old revisions, so it does not matter.
-
Which thing exactly? It probably isn't doing that thing for the old revisions, so it does not matter.
upload-pack. It's a low-level git command.
-
Rewriting published-and-depended-upon history is a recipe for trouble
Yes, it was clearly in the danger zone, but I didn't exactly rewrite what was already pushed to upstream (there was no push -f). I just left a detached (with no branch label) commit in my local history, and pointed to it from a parent repo. That of course breaks the submodule for that particular commit in the parent repo, but what's interesting is that it can't be fixed with simply pointing the submodule to a good (existing in upstream) commit hash in the submodule, commiting and pushing. It's like gitlab tried to recreate all previous states.
-
Something I have noticed happening more and more is that when my password manager fills my username and password for me on some websites and I hit Tab and then Enter or Space I get an error similar to "Please enter a password". I don't know enough about web development to know how something that retarded could happen but I assume that they are checking for keystrokes in the box and not just checking whether there is information in the box and that seems really fucking retarded.
My bank has also fucked up their Android app so that occasionally, but not all the time, password manager will input the password into the username and password box both. How you fuck something up like that and get it to only happen on some loads is a mystery that boggles my mind.
On a related note I find it really annoying that on some websites that are SPAs Space doesn't work to submit. IIRC, Google Chat or whatever the fuck they are calling it now is like that and it really fucking annoys me. Another less common anti-pattern but more annoying than all the rest combined is when a page is laid out so that the
Cancelbutton comes before theSubmitbutton and then muscle memory causes me to clear all the shit I just typed in.
-
@sebastian-galczynski said in WTF Bites:
Which thing exactly? It probably isn't doing that thing for the old revisions, so it does not matter.
upload-pack. It's a low-level git command.upload-packis the server-side bit called during fetch. The important question is, why was it called? What was the client-side that got the error? I can only think about something trying to checkout the revision of the parent repository with referencing the missing commit, but if you pushed a revision with a good ref on top of that, I'd expect it to stop being a problem.
-
@Polygeekery said in WTF Bites:
web development
@Polygeekery said in WTF Bites:
that seems really fucking retarded
Other retarded login forms that cause problems with password managers are the ones that prevent pasting into the password box.
-
@Polygeekery said in WTF Bites:
I don't know enough about web development to know how something that retarded could happen
Have you met today's web developers?
-
@loopback0 said in WTF Bites:
Other retarded login forms that cause problems with password managers are the ones that prevent pasting into the password box.
And the forms that are laid out in such a fucked up manner that even though they are asking for your address or email address or other information that is stored in your browser or password manager or wherever they are unable to discern that and no auto-populate options come up. Another gigantic fuckup is when they are laid out more fucked up and an off-by-X manner and everything gets populated in the wrong location.
Why are so many web developers so fucking retarded?
-
@TimeBandit said in WTF Bites:
@Polygeekery said in WTF Bites:
I don't know enough about web development to know how something that retarded could happen
Have you met today's web developers?
Yes. But the ones that I know do not require velcro shoes.
-
@Polygeekery said in WTF Bites:
Why are so many web developers so fucking retarded?
In this particular case it's the security “experts” that tend to be so fucking retarded—the usual argument for weird login dialogs that break autofill and such is
security.
-
@Bulb yeah, breaking auto fill is a “feature”, because these
security experts are certified morons. Probably with expensive, physical certificates.
That’s mostly banks etc. that go out of their way to have the worst security. But even shit like google, who are the “web natives” that should know better, have questionable designs. What’s up with bullshit two step pages where I first enter user name, submit, then password, submit. Is that supposed to break auto fill? (It used to, but works by now) Or does that have another stupid rationale?Also, goddamn phones: I know that password entry fields are supposed to not show the text, but that’s pretty fucking stupid on a phone. Give me a way to turn this off.
-
@Polygeekery said in WTF Bites:
Why are so many web developers so fucking retarded?
You and your rhetorical questions.
-
Also, goddamn phones: I know that password entry fields are supposed to not show the text, but that’s pretty fucking stupid on a phone. Give me a way to turn this off.
On Android this can be turned on and off.
-
In this particular case it's the security “experts” that tend to be so fucking retarded—the usual argument for weird login dialogs that break autofill and such is security.
The guy running the Begraben project (which is now indeed buried) demanded that we do this. This resulted in some kind of arms race with Mozilla, with each version of Firefox breaking our breakage of password auto-fill. I think at the end the password field didn't have type=password, and the word 'password' in the label had to be obfuscated with s, or something like that.
-
@sebastian-galczynski said in WTF Bites:
Also, goddamn phones: I know that password entry fields are supposed to not show the text, but that’s pretty fucking stupid on a phone. Give me a way to turn this off.
On Android this can be turned on and off.
Yeah, on my ancient Android phone, password fields (always? sometimes?) have a little "eye" icon I can tap to reveal the contents of the field. I find this mostly useful with KeePass, because my master passphrase is long and easy to typo; everything else gets copy-pastad from there.
-
What’s up with bullshit two step pages where I first enter user name, submit, then password, submit. Is that supposed to break auto fill? (It used to, but works by now) Or does that have another stupid rationale?
There are a couple different reasons. Stupid? I dunno.
Why are the username and password on two different pages?
Twilio posts cloud communications trends, customer stories, and tips for building scalable voice and SMS applications with Twilio's APIs.
-
The two-page design also makes it harder for bad actors to create phishing sites with look-alike login pages when there is a page redirect involved.
-
@TimeBandit said in WTF Bites:
The two-page design also makes it harder for bad actors to create phishing sites with look-alike login pages when there is a page redirect involved.
You have to fake twice as many pages, so QED.
-
@loopback0 said in WTF Bites:
Other retarded login forms that cause problems with password managers are the ones that prevent pasting into the password box.
Morons like this are why the Don't Fuck With Paste extension exists.
-
What’s up with bullshit two step pages where I first enter user name, submit, then password, submit. Is that supposed to break auto fill?
No, that's because of pluggable authentication. They don't know whether there even will be password before you tell them the username.
I am not sure they got around to implementing any passwordless authentication methods yet, but at least Microsoft allows companies to set up delegating the login to their own systems. So for some companies you enter the login and then (depending on the domain) instead of proceeding to password it redirects you to a different login screen from a different server.
-
@TimeBandit said in WTF Bites:
The two-page design also makes it harder for bad actors to create phishing sites with look-alike login pages when there is a page redirect involved.
Yeah, that's bullshit. The redirects make it less obvious for user to tell what is legitimate and what isn't.
-
Also, goddamn phones: I know that password entry fields are supposed to not show the text, but that’s pretty fucking stupid on a phone. Give me a way to turn this off.
This reminds me of a certain CLI tool which, rather than turning terminal echo off, would generate a backspace-space-backspace sequence whenever it received a character.
-
Status: Publishing licenses sounds like a parody of themselves, but they do not mess around (bold in original, bold-italics mine):
In consideration for publication in the above Journal, you hereby grant [bla bla bla ...] the exclusive right and licence to produce, publish and make available and to further sub-license your article ('Article') and the accompanying abstract (all materials collectively referenced as the 'Contribution') prepared by you for the full legal term of copyright and any renewals thereof throughout the universe in all languages and in all formats, and through any medium of communication now known or later conceived or developed.
I'm sure the Greys on Alpha Centauri care a whole lot about our shit.
-
@topspin I've seen that before.
I hate that traditional style of publishing licenses. You're screwed either way. Thankfully, there's a lot more open agreements around these days (including various CC-based ones). Though, somewhat painfully, if you're not covered by an existing agreement, these might set you back a few €1000.
(I still think we should burn the old-style publishing to the ground. One might ask "what about journal rankings? How do I get my citation scores?". We'll find a way if nobody can publish in those established places.)
-
@cvi it's all bullshit one way or the other, so, yes.
-
On a related note I find it really annoying that on some websites that are SPAs Space doesn't work to submit.
There seems to be a bizarre shift towards styling a div to look like button rather than actually... using a button. Surprised you can tab on to it. It's not hard to do but people looooooove making pretty but useless widgets.
-
@sebastian-galczynski said in WTF Bites:
obfuscated with s, or something like that.
Oh, nice html injection you have here. I wrote
obfuscated with <span>'s, and the tag is still there. Unescaped.let's see...
dupa
Ok, style is removed on display.
-
@sebastian-galczynski said in WTF Bites:
Oh, nice html injection you have here.
Some HTML tags are intentionally supported. It's easier than make up a
syntax for the same thing. I Hope™ they are actually parsed and printed back. Note that <span>will render as actual span, but< span >will render as text: < span >.
-
Tip: next time when someone, in a discussion about integration of legacy system, tries to argue that "such a mess would not be possible in real life", show them this:
Stavbaři při přístavbě domu obezdili elektrické vedení. Neskutečné, energetici kroutí hlavou
Podle společnosti ČEZ investor nepostupoval v souladu s podmínkami distributora a patrně i v rozporu s podmínkami stanovenými stavebním povolením.
(sadly, no english source with good pictures... but no comment is necessary anyway).
-
@Kamil-Podlesak
, somewhere, I think by @blek.
-
@Polygeekery said in WTF Bites:
assume that they are checking for keystrokes in the box and not just checking whether there is information in the box and that seems really fucking retarded.
If memory serves, it is because the data isn't actually in the text field until after you leave the field, and the "blur" event that actually has the ghost-written password happens after the "key up" event during which the text isn't actually there.
And then some sites try to block auto-fill by clearing the textbox a few times depending on which box you "clicked" in first (thanks BoA).
-
@Bulb Yeah, but it was in the Garage.
-
-
@Bulb Yeah, but it was in the Garage.
Understandable, but those are not trolleys.
-
Thankfully, there's a lot more open agreements around these days (including various CC-based ones). Though, somewhat painfully, if you're not covered by an existing agreement, these might set you back a few €1000.
I did end up going with CC BY 4.0.
That appparently would've cost a mere £2750, which is dirt cheap, considering the journal gets the original research, the peer reviews, and (probably?) the editor for free, and publishing is electronic. Fortunately, there was an "arrangement with German Academic insitutions" which reduced that cost to £0.
-
@topspin I'm considering to start selecting co-authors based on whether their affiliations cover publishing fees in the target journals.
-
The controversial leap second, which time keepers add sporadically to keep atomic clocks aligned with Earth’s rotation, will be axed in 2035, the International Bureau of Weights and Measures (BIPM) decided on 18 November. Devised in 1972 and used 27 times since, the leap second wreaks havoc with modern-day telecommunications, banking, and other networks. Its abandonment means that astronomical time, based on Earth’s rotation, will slowly diverge from Coordinated Universal Time, based on the vibrations of cesium in atomic clocks. BIPM plans to stop adding leap seconds for 100 years, by which time someone may have figured out a long-term fix for the problem.
...because as we all know,
// TODO: fix this lateris a very effective way to solve problems.
-
@Zerosquare said in WTF Bites:
based on Earth’s rotation
Found the problem. This is the kind of Shit that spawned daylight savings time, after all.
-
@Tsaukpaetra a reasonable time standard would be the real value between 0.0 (Sol ignition) and 1.0 (Sol exhaustion). Anything smaller is titchy parochialism.
-
@Gribnit That sounds like a serious outbreak of heliocentricism. Please see your local health care professional regarding that.
