WTF Bites
-
Youtube recently started listing videos in some random-ass order in a channel's video list. That was driving me up the wall; I didn't notice until now that it had the following buttons (maybe they added it later):
The default is "For you" and I just want it to stick to "Latest".
Googling this is particularly unhelpful. Might be that I'm being A/B:d. But if anybody happens to know how to change the default for that setting ...
... oh, yeah, and
Youtube.
-
-
Youtube recently started listing videos in some random-ass order in a channel's video list. That was driving me up the wall; I didn't notice until now that it had the following buttons (maybe they added it later):
The default is "For you" and I just want it to stick to "Latest".
Googling this is particularly unhelpful. Might be that I'm being A/B:d. But if anybody happens to know how to change the default for that setting ...
... oh, yeah, and
Youtube.I finally quit my YT Premium subscription because every single update made shuffle playlists break a little more. And when I tried opening a support case for it with detailed reproduction steps specifying that my primary use was via Chrome on desktop, they were like “lol update the app?”
-
@loopback0 said in WTF Bites:
Might be that I'm being A/B:d
I think so - hasn't hit me yet
Yeah, I'm just seeing “Latest” and “Popular”, no “For you” nonsensium.
-
@loopback0 said in WTF Bites:
Might be that I'm being A/B:d
I think so - hasn't hit me yet
Yeah, I'm just seeing “Latest” and “Popular”, no “For you” nonsensium.
I scrolled to the right, and there's a send feedback.
-
@Tsaukpaetra said in WTF Bites:
@loopback0 said in WTF Bites:
Might be that I'm being A/B:d
I think so - hasn't hit me yet
Yeah, I'm just seeing “Latest” and “Popular”, no “For you” nonsensium.
I scrolled to the right, and there's a send feedback.
The over shareing your horrific taste in youtube subscriptions thread is
-
-
Youtube recently started listing videos in some random-ass order in a channel's video list. That was driving me up the wall; I didn't notice until now that it had the following buttons (maybe they added it later):
The default is "For you" and I just want it to stick to "Latest".
Googling this is particularly unhelpful. Might be that I'm being A/B:d. But if anybody happens to know how to change the default for that setting ...
... oh, yeah, and
Youtube.I finally quit my YT Premium subscription because every single update made shuffle playlists break a little more. And when I tried opening a support case for it with detailed reproduction steps specifying that my primary use was via Chrome on desktop, they were like “lol update the app?”
Well hey, now I'm getting the "Latest" and "Popular" tags, which also means that they've removed the automatic playlists that you used to be able to get to play all videos by a particular uploader without them having to curate playlists.
Definitely going to reinvest that YT++ money into
soy double half-calf mocha lattesmoar online gaming subs.
-
@Tsaukpaetra said in WTF Bites:
youtube subscriptions
I swear it's like Google doesn't even know me....!
-
@Applied-Mediocrity said in WTF Bites:
@Tsaukpaetra said in WTF Bites:
youtube subscriptions
I swear it's like Google doesn't even know me....!
We're missing the last two panels!
-
-
@Applied-Mediocrity said in WTF Bites:
@Tsaukpaetra said in WTF Bites:
We're missing the last two panels!
Show, don't tell.
To the NSFW thread!
-
Status: This computer is running a RAID0 with department-critical data.
What the hell....
-
It fully deserves your stamp of approval.
-
WTF of my day: So, we're running BigBlueButton as a video conferencing tool at my school. It worked quite nicely so far even if the underlying architecture is Holy-Jumble-Of-Everything-Up-To-And-Including-The-Kitchensink-Batman.
I mean, seriously, there's Meteor as a framework (and thus Node.js), then we've got RoR, Java, something is pulled from Docker and various other services...
... but it does run.
Then I was notified that there's a new version out there (2.6) which is adding quite a number of attractive tools (like being able to force a layout on every client - you thus can switch e.g. from presenter view to presentation view and not have tell everyone to maximize the presentation).
Took a look at the upgrade instructions and noped out. And instead opted for a clean install which is made very easy - you just have to run one script and it does everything for you, given the proper arguments. Plus, our two servers are living behind a load balancer anyway so that part I didn't even have to touch.
So, upgrading basically went:
- wipe the server with a minimal Ubuntu 20.04 install
- start the install script
- get the server secret and paste that into the load balancer's config so it can actually talk to the instance
But then we get into the territory of connections - y'see, our administration's PCs are living behind a very restrictive firewall. And thus a TURN/STUN server is needed to yeet the WebRTC connections over the 443 port.
The manual explicitly states that such a server should exclusively hold this service on its own. That's why we got a cheap VirtualRoot server just for that. Which was working fine.
While going over the install script to see which arguments needed replacing for the proper values I noticed a new argument - namely providing the address and the secret for the STUN server. "Neat!" I thought and plonked the proper values in there.
So, install done, everything worked ... except for the connection from the administrative net (which did work before!). Examined the WebRTC logs and noticed that they were trying to connect to the wrong STUN server. Hmmh, seemingly a typo. But no problem, edited the configuration file referenced in the manual for this purpose and ... still the wrong server?
?I then tried to find out where the wrong value was coming from - no dice. But okay, since reinstalling was so fast and I had misconfigured in the beginning, I'd simply reset to start.
So, blank slate and then the install script again - this time without any STUN arguments, though, just to see the difference.And then I saw it: The script installed a STUN server! On the same instance!
Tried to connect from the administrative net and it does work.
But seriously: This is very annoying - the manual still explicitly states that you need your own STUN server, that you need to configure it in that way and then the new version simply pulls the value from somewhere else and you don't need it anyway...
-
-
So, we're running BigBlueButton
I understand the need for anonymization, but did you have to pick such a ridiculous name?
-
WTF of my day: So, we're running BigBlueButton
The rest of your post is effectively redundant.
-
So, we're running BigBlueButton
I understand the need for anonymization, but did you have to pick such a ridiculous name?
Fred liked it, but it really is called that.
-
But seriously: This is very annoying - the manual still explicitly states that you need your own STUN server, that you need to configure it in that way and then the new version simply pulls the value from somewhere else and you don't need it anyway...
Stunning!
-
-
Apparently, when LibreOffice updates itself, it discards any history of recently opened documents. With it goes the ability to recover any documents that were open the last time the system was involuntarily rebooted.
-
Twitter was having problems with its reply for a couple of hours.
This was the error I was getting in the console:
Uncaught (in promise) Error: ApiError: https://api.twitter.com/graphql/C2dcvh7H69JALtomErxWlA/CheckTweetForNudge HTTP-200 codes:[37] trace_ids:["d74d6fa8e0aa1318"]
at new a (vendor.4cfde47a.js:82)
at d.<anonymous> (vendor.4cfde47a.js:82)
at new d (main.c36fee8a.js:1)
at d.<anonymous> (vendor.4cfde47a.js:82)
at new d (main.c36fee8a.js:1)
at main.c36fee8a.js:1
(anonymous) @
-
@HardwareGeek said in WTF Bites:
Apparently, when LibreOffice updates itself,
welltheresyourproblem.jpg
-
I'm pretty sure Teams doesn't show messages until I "come back" to the "idle computer" until I move the mouse or make some other sort of input. This includes when Teams is the foreground application and the messages are in the currently visible channel.
-
I've seen other applications acting in the same manner. I think they're delaying their UI refreshes until they get a "computer is no longer idle" signal. Which is probably a good thing to avoid wasting energy. (Yeah, I know, all the other lazy design choices more than offset the gain.)
-
@Zerosquare hmm...yeah, might be a consequence of chrome's new memory saver thing?
-
Fred
I understand the need for anonymization, but couldn't you have picked a more ridiculous name?
-
WTF of my day: From a ticket reported by the manager of a product that we acquired and are "supporting" (it's frozen, but still some customers paying lots of money):
The problem now reported is that the 6-digit incident ID # is not actually unique because the dispatch numbering system restarts with the new calendar year. As a result, when a dispatch for [agency] in 2023 has the same 6-digit incident # as a 2022 dispatch for [agency], the XML files generated will be written as updates for the 2022 dispatch...
-
@Benjamin-Hall said in WTF Bites:
WTF of my day: From a ticket reported by the manager of a product that we acquired and are "supporting" (it's frozen, but still some customers paying lots of money):
The problem now reported is that the 6-digit incident ID # is not actually unique because the dispatch numbering system restarts with the new calendar year. As a result, when a dispatch for [agency] in 2023 has the same 6-digit incident # as a 2022 dispatch for [agency],
So all you need to do is archive all last year's files, starting from the start of last year, starting before the end of the year - what could go wrong?
the XML files generated will be written as updates for the 2022 dispatch...
At least it's XML.
-
Fred
I understand the need for anonymization, but couldn't you have picked a more ridiculous name?
That’s the comedy, I’m actually not.
He says it himself. Nice guy, is Fred.
-
I'm pretty sure Teams doesn't show messages until I "come back" to the "idle computer" until I move the mouse or make some other sort of input. This includes when Teams is the foreground application and the messages are in the currently visible channel.
Speaking of Teams, it seems that when someone reacts to a chat message (with e.g.
), the chat is marked as "changed" (i.e. bolded) but if it is the active one the status never updates. When a new message is typed it usually ends up noticing that yes, I have seen that message, but for reactions, nope.I even tried switching to a different chat and back, doesn't seem to work. The only thing that reliably works to get rid of the notification is to go into the "Activity" feed.
-
WTF of my day: I'm getting a bit annoyed with one of the services we're running. Basically, the service dies whenever I do
apt update && apt upgrade.Why? Because if said service receives a package update it will insist on running a DB migration (even if the schema didn't change) and simply crash out until then. Now, I'd understand the demand for manual intervention if you had any choice about it. But, no, you have to do a
foo-service --configure, it does any and all migrations without any kind of interactivity and then it's off to the races again.So I'm not quite getting why this isn't done via a post-install hook automatically - it's not as if I can choose to not run
foo-service --configureafter I already upgraded the packages. It's mandatory at this point. Probably a case of package maintainer and the person responsible for the script not talking to each other.Oh, and after doing that you better hope that you have a backup of your configuration file - because that script will also nuke said file from orbit and install the default one (without asking, of course).
If only
apthad a way of dealing with configuration file changes...
-
@Rhywden which particular half-assed educational software written by open source geniuses is giving you that grief? Because no-one else gets it RIGHT that hard like the edu adjacent crowd.
-
@Rhywden which particular half-assed educational software written by open source geniuses is giving you that grief? Because no-one else gets it RIGHT that hard like the edu adjacent crowd.
OpenProject
It using RoR under the hood. Go figure.
-
@Rhywden I mean, I was absolutely right about it being open source geniuses, and you have RoR on you, quelle surprise. But I have to admit I have not played with OpenProject - now it may fulfil the sufficient level of WTF to be on my list of disgusting curiosities that need a dissection for morbid fascination purposes.
-
WTF of my day: I'm getting a bit annoyed with one of the services we're running. Basically, the service dies whenever I do
apt update && apt upgrade.Why? Because if said service receives a package update it will insist on running a DB migration (even if the schema didn't change) and simply crash out until then. Now, I'd understand the demand for manual intervention if you had any choice about it. But, no, you have to do a
foo-service --configure, it does any and all migrations without any kind of interactivity and then it's off to the races again.So I'm not quite getting why this isn't done via a post-install hook automatically - it's not as if I can choose to not run
foo-service --configureafter I already upgraded the packages. It's mandatory at this point. Probably a case of package maintainer and the person responsible for the script not talking to each other.Oh, and after doing that you better hope that you have a backup of your configuration file - because that script will also nuke said file from orbit and install the default one (without asking, of course).
If only
apthad a way of dealing with configuration file changes...That sounds like some things I've worked with in the past (lab info management software is at least as terrible as education software). They also included the magnificent decision to require multiple services all on the root of port 80, but which couldn't be on the same machine and which needed to do full cross site scripting. I ended up hacking around with VMs and nginx and got it all working, but it was ugly as sin.
You know it is bad when you have one "service" that needs PHP, Java and C# all at the same time.
-
now it may fulfil the sufficient level of WTF to be on my list of disgusting curiosities that need a dissection for morbid fascination purposes
Please report here!
-
@dkf Yeah, some of that stuff was dreamt up by someone who saw a Rube-Goldberg-machine and was way too inspired by that.
Keycloak being another example. I recently looked at upgrading our outdated install (v14 I think) to a more recent one and, uh, rebuilding from scratch is probably way faster.
-
@Rhywden cromulent with my experiences of Keycloak.
-
@Arantor I'm not even beginning to understand why it's that complicated under the hood.
-
So I'm not quite getting why this isn't done via a post-install hook automatically - it's not as if I can choose to not run
foo-service --configureafter I already upgraded the packages. It's mandatory at this point. Probably a case of package maintainer and the person responsible for the script not talking to each other.Well, it's not an official Debian or Ubuntu package, is it? In that case the most probably cause is that whoever makes the package uses some quick-and-dirty-slap-together-a-dpkg script (and similar one for rpm and possibly a handful more formats too) and does not even know there is such a thing as post-install script.
-
@Bulb In that case the person is sleeping on the ... uh, job. I knew that such things exist and that only by looking at what happens after
apt upgradeeven on a default Ubuntu installation with not much on it.Not to mention neglecting to even beginning to think about such things as: "Well, there must be a way to run such stuff after we installed our package." but well...
-
In that case the person is sleeping on the ... uh, job
There seems to be a pandemic of sleeping on the job going on (for quite a while already, but it's getting progressively worse).
-
the person is sleeping on the ... uh, job.
Dude, I have small kids. When else am I supposed to rest?
-
@dkf I'm currently considering whether to use RabbitMQ for something because their
Streamsplugin does a lot of what I'd like to have (basically, it's a streamed queue where every client can go back to an arbitrary point in the queue to fetch, say, messages which arrived after the client was last online).Only real problem: This feature uses TCP connections. Which means that any port I open to connect to is solely in the possession of this feature. Usually I could say: "Okay, request at port 80 for www.baz.de goes to this server in the back and request for port 80 at www.foo.de goes there" but a TCP connection does not provide which URL it tried to go to.
And I just know that if I use a port like 5225 for this it'll be blocked at some firewalls just because.
-
@Rhywden Can it stream over a websocket?
-
There seems to be a pandemic of sleeping on the job going on
Yeah, but I'll be having my afternoon caffeine fix soon.
-
@dkf I'm currently considering whether to use RabbitMQ for something because their
Streamsplugin does a lot of what I'd like to have (basically, it's a streamed queue where every client can go back to an arbitrary point in the queue to fetch, say, messages which arrived after the client was last online).That sounds more like the Kafka's shtick, but it should work in Rabbit too, yes.
Only real problem: This feature uses TCP connections. Which means that any port I open to connect to is solely in the possession of this feature. Usually I could say: "Okay, request at port 80 for www.baz.de goes to this server in the back and request for port 80 at www.foo.de goes there" but a TCP connection does not provide which URL it tried to go to.
A TLS connection does possess a SNI though so for amqps you can do multiplexing—with sufficiently modern proxy that can do SNI-based routing.
And I just know that if I use a port like 5225 for this it'll be blocked at some firewalls just because.
It's port 5672 for clear-text and 5671 over TLS.
Yes, it might be blocked at some corporate firewalls, and it won't be available from thin (web) clients.
Whether it's a problem depends on what your use-case is.
@Rhywden Can it stream over a websocket?
No, at least not easily. It does not support AMQP over websockets, only MQTT over websockets, but I don't think MQTT supports streams.
The new cool kid on the block of messaging is NATS, and that one does advertise websocket support for all its features. Unlike rabbit, I didn't see it actually in production, but it looks like it might suit you better (and it's more efficient; erlang is good for reliability and scaling, but efficient it isn't).
-
@dkf No, seems to be a binary protocol. The other ones like STOMP do have a websocket plugin, this does not.
@Bulb No,
Streamis on port 5552 by default - it's not AMQP but a bit different.
x.com
