WTF Bites
-
@levicki said in WTF Bites:
Pointless
is pointless.Except it's not pointless when it impacts your ability to read other people's posts. Everyone else, when they say Unicode, they mean Unicode as a whole and all its possible encodings, it's just you who's the odd one out and insists on using wrong terminology.
Duh? Do you also complain when people do math tricks that only work on little endian numbers?
Of course, they need to work everywhere.
Sure, and it's not a good assembly code if it doesn't run on both x86 and ARM.
Your challenge, literally as you wrote it.
Nope.
Yep. If you disagree, show me which part I've got wrong.
As portable as can be. UTF8-CPP is very small header-only library that works with every compiler.
So you didn't solve it then!
I did. Using 3rd party libraries for things that were done 1000 times over is a valid solution. I'm a software developer, not surreal quiz solver.
The requirement was to use only standard libraries that come with the language
I explicitly said that most languages don't natively support it very well
I explicitly said that 3rd party libraries help with that as long as the encoding-agnostic part is done right, and most of the time it is.
not that it is impossible to find or write a library that handles it.
If you know there are libraries that solve exactly the problem you're having, then what's the problem?
If they want better adoption they should make it part of the standard library of every language and make it easy to use.
I don't think they really crave better adoption. I think they're fine with 99.999% of new projects.
-
That has to come sometime after UTF-95...
And it's probably followed by UTF-1.
-
@levicki said in WTF Bites:
If you know there are libraries that solve exactly the problem you're having, then what's the problem?
The problem is with you complaining that people don't want to use Unicode. I am just telling you why (because native support sucks)
Let me repeat: if you know there are libraries that solve exactly the problem you're having, then what's the problem?
and you want to shoot the messenger
You're very damn right I want to shoot every messenger who preaches laziness and apathy and recommends using vastly inferior technology that causes billions of compatibility problems. That's how you get rid of the problem - by shooting people that cause it.
not to mention you are essentially saying "it is easy if you just use unvetted 3rd party code that deals with unsanitized input in all your projects".
You can vet if you want. It's just 900 lines.
-
You can vet if you want. It's just 900 lines.
I bookmarked it, by the way. It indeed looks pretty short.
Not that I think I'll ever use it (or have even read the README to figure out its exact purpose), but you never know.
-
@topspin I think Blakeyrat linked that thread as an example of how every thread on WTDWTF eventually ends up being about C and/or C++ language details. Because it did not take long for people to start debating that in that one...
Message from the beyond:
-
@Tsaukpaetra What's my avatar doing in that screenshot?
e: Oh, it's my post he's quoting
-
@Tsaukpaetra What's my avatar doing in that screenshot?
e: Oh, it's my post he's quoting
Yeah, NodeBB helps in mysterious ways.
-
I think Blakeyrat linked that thread as an example of how every thread on WTDWTF eventually ends up being about C and/or C++ language details.
Another blakeyrat law?
-
-
@boomzilla said in WTF Bites:
@Tsaukpaetra said in WTF Bites:
Message from the beyond:
That guy is such a retard.
"echo back to the beyond..."
-
@loopback0 said in WTF Bites:
Another blakeyrat law?
Blakeyrat's Law #(rand): Every thread on WTDWTF eventually devolves into a Blakeyrant.
-
@boomzilla said in WTF Bites:
@Tsaukpaetra said in WTF Bites:
Message from the beyond:
That guy is such a retard.
And here's Boomzilla to tell me all my ideas are stupid!@blakeyrat said on DiscourdBB:
This thread is a good example of how
if ANYBODY brings up C or C++ the entire thread turns into a deep-dive into the C language specification. Every time.people on a programming forum always find a way to talk about such boring shit as programming.
-
@boomzilla said in WTF Bites:
@Tsaukpaetra said in WTF Bites:
Message from the beyond:
That guy is such a retard.
And here's Boomzilla to tell me all my ideas are stupid!@blakeyrat said on DiscourdBB:
This thread is a good example of how
if ANYBODY brings up C or C++ the entire thread turns into a deep-dive into the C language specification. Every time.people on a programming forum always find a way to talk about such boring shit as programming.NO ONE ASKED FOR BLAKEYRAT TO COMMENT ON ANYTHING
-
@levicki said in WTF Bites:
You can vet if you want. It's just 900 lines.
Yes, it's just 900 lines for UTF-8, it's another few hundred for RapidJson, another few hundred for TinyXML, then for whatever else standard library is still lacking, and which is nowadays considered essential for development.
Sure, why not waste time vetting heaps of someone else's code instead of using that time to write the actual code you need?
Because despite all the effort and expertise required, vetting will still be faster? Also, it'll most likely have more features and less bugs - if only because it was already tried and tested by others before you. Note that I'm saying "likely" - I don't say this will be the case every time.
Do you write your own compilers too? Because, you know, compilers are a lot of lines to vet.
Not sure about you but I'd rather have a standardized way of using UTF-8 in the language runtime -- coded, vetted, and vulnerability tested by professionals, not to mention being kept up to date and in sync with the toolchain.
Here you are:
ICU - International Components for Unicode
News 2024-04-17: ICU 75 is now available. It updates to CLDR 45 (beta blog) locale data with new locales and various additions and corrections. C++ code now requires C++17 and is being made more robust. The CLDR MessageFormat 2.0 specification is now in technology preview, together with a
The official libraries for C, C++ and Java developed by the Unicode Consortium themselves. Doesn't get any more professional than that.
-
Back on topic.
I didn't even start using Visual Paradigm 16 and I already hate it. Compared to version 15, it starts like ten times slower, and shows an always-on-top splash screen that makes me unable to do anything else in the meantime. Ugh.
On first startup, it shows "what's new" window. I clicked through a few slides, and it's all stuff that sounds very advanced and specific and I will have zero use of. But after closing it, it shows this beautiful dialog box:
FUCK YOU!!!!!!!!
-
Did I ever mention the weird positive feedback loop you get when you upload screenshots on a Windows 10 computer with display scaling >100%? Basically, the screenshots are from on-screen pixels, so it's already scaled up in the image - and when displaying the screenshot, the scaling gets applied to the picture, effectively scaling the window twice. If I made a screenshot of that screenshot, it would get even bigger.
I know there's no practical way to do it differently, but it still annoys me. What I see is not what I get.
-
@levicki said in WTF Bites:
Yes I am aware that Unicode encompasses UTF-8 encoding, but for most people, especially those with Windows background "Unicode" means wide strings (UTF-16 / UCS-2) instead of MBCS such as UTF-8.
It really doesn't. "Unicode" means UTF-8 pretty much all the time, unless you're Microsoft.
-
@anonymous234 also, Java. Also also, JavaScript.
-
@anonymous234 said in WTF Bites:
"Unicode" means UTF-8 pretty much all the time
No. Unicode is a charset that describes a bunch of characters and a mapping between those and numbers (in the range up one million or so, quite a lot bigger than the charsets that preceded it). UTF-8 is a scheme for encoding a sequence of Unicode characters as bytes; there are others too.
Programs may or may not use UTF-8 internally. As long as they don't screw up the Unicode character sequence, the encoding is of little interest (you wouldn't believe how angry this fact made certain junior employees of the Unicode Consortium) and there are solid programming reasons for wanting something else (in particular, non-constant-width encodings increase the complexity class of a lot of algorithms). For data at rest or being exchanged between programs, UTF-8 is absolutely superior.
-
@levicki said in WTF Bites:
Notepad++ developer decided to make a statement about Chinese oppression of Uyghur people and has released Notepad++ Free Uyghur Edition.
Did it backfire? Absolutely. For starters, GitHub project was vandalized and he received a bunch of emails full of insults from angry Chinese developers.
He doesn't seem to give up though.
This should not come as a surprise.
Here's another controversial release: https://notepad-plus-plus.org/news/v674-je-suis-charlie-edition/
And its corresponding report after the website got defaced: https://notepad-plus-plus.org/news/npp-site-hacked-cos-je-suis-charlie-edition/
-
@levicki said in WTF Bites:
I am not saying he shouldn't try if he wants, it's his program, but to me it looks kinda pointless.
It is - and also I want political lessons from a text editor about as much as I want cooking tips from Visual Studio.
-
@loopback0 said in WTF Bites:
I want cooking tips from Visual Studio.
Hmm....
Nope, still don't know what to cook for dinner.
-
Did I ever mention the weird positive feedback loop you get when you upload screenshots on a Windows 10 computer with display scaling >100%? Basically, the screenshots are from on-screen pixels, so it's already scaled up in the image - and when displaying the screenshot, the scaling gets applied to the picture, effectively scaling the window twice. If I made a screenshot of that screenshot, it would get even bigger.
I know there's no practical way to do it differently, but it still annoys me. What I see is not what I get.
Do any image formats encode DPI so that the browser's automatic image sizing code can account for it?
-
Did I ever mention the weird positive feedback loop you get when you upload screenshots on a Windows 10 computer with display scaling >100%? Basically, the screenshots are from on-screen pixels, so it's already scaled up in the image - and when displaying the screenshot, the scaling gets applied to the picture, effectively scaling the window twice. If I made a screenshot of that screenshot, it would get even bigger.
I know there's no practical way to do it differently, but it still annoys me. What I see is not what I get.
Do any image formats encode DPI so that the browser's automatic image sizing code can account for it?
Exif tags Exif.Image.XResolution and Exif.Image.YResolution. Been around since forever.
-
Did I ever mention the weird positive feedback loop you get when you upload screenshots on a Windows 10 computer with display scaling >100%? Basically, the screenshots are from on-screen pixels, so it's already scaled up in the image - and when displaying the screenshot, the scaling gets applied to the picture, effectively scaling the window twice. If I made a screenshot of that screenshot, it would get even bigger.
I know there's no practical way to do it differently, but it still annoys me. What I see is not what I get.
Do any image formats encode DPI so that the browser's automatic image sizing code can account for it?
Exif tags Exif.Image.XResolution and Exif.Image.YResolution. Been around since forever.
PNG, Windows BMP, and Windows ICO have fields specifying pixels-per-metre.
-
If they own an ancient medium using ancient data formats, they surely have an ancient device to play it with as well.
Surely not. I have 5.25" floppies with files in Word for DOS. I do not have a 5.25" floppy drive with which to (try to; it's likely bit-rotted to unreadability) read them.
-
Your right-to-bear-
-on-your-belt has just been revoked.
-
@levicki: The NSFW thread is
-
@HardwareGeek said in WTF Bites:
I have 5.25" floppies
withthat used to have files in Word for DOS last time I checked.FTFY
-
Did I ever mention the weird positive feedback loop you get when you upload screenshots on a Windows 10 computer with display scaling >100%? Basically, the screenshots are from on-screen pixels, so it's already scaled up in the image - and when displaying the screenshot, the scaling gets applied to the picture, effectively scaling the window twice. If I made a screenshot of that screenshot, it would get even bigger.
I know there's no practical way to do it differently, but it still annoys me. What I see is not what I get.
Do any image formats encode DPI so that the browser's automatic image sizing code can account for it?
Exif tags Exif.Image.XResolution and Exif.Image.YResolution. Been around since forever.
Did I ever mention the weird positive feedback loop you get when you upload screenshots on a Windows 10 computer with display scaling >100%? Basically, the screenshots are from on-screen pixels, so it's already scaled up in the image - and when displaying the screenshot, the scaling gets applied to the picture, effectively scaling the window twice. If I made a screenshot of that screenshot, it would get even bigger.
I know there's no practical way to do it differently, but it still annoys me. What I see is not what I get.
Do any image formats encode DPI so that the browser's automatic image sizing code can account for it?
Exif tags Exif.Image.XResolution and Exif.Image.YResolution. Been around since forever.
PNG, Windows BMP, and Windows ICO have fields specifying pixels-per-metre.
Alas, it was not meant to be... https://stackoverflow.com/a/27328679/1959975
The DPI value stored in an image can be used to scale it automatically to the correct physical size when it's displayed on a device whose DPI is also known. For example, a 144dpi image displayed on a 72dpi monitor should be scaled by 50% in each dimension, so that 144 pixels (one inch) of image is mapped to 72 pixels (one inch) of monitor surface. Web browsers, however, typically don't do this; image pixels are simply mapped directly to screen pixels, so you have to manually scale your images to have pixel dimensions that are appropriate for the monitor where they'll be displayed.
Thanks web browsers for doing it wrong! I mean, maybe there is some good reason they don't do it. But I can't think of what it would be. At least now I have learned about
window.devicePixelRatioand can add "make userscript to fix image DPI in browsers" to my "want to do this but will never find time for it" list.
-
@LB_ the problem is, since no browser (and probably no other image viewing application) supports this, the images themselves likely never have it set correctly. So your endeavor would be quite pointless.
-
Thanks web browsers for doing it wrong! I mean, maybe there is some good reason they don't do it.
It seems annoyingly hard to determine from inside the browser that, e.g., the two monitors in front of me here have display sizes of 52×29 and 30×47 cm². Fortunately I don't have any additional scaling factors; projectors are a whole 'nother issue again.
-
But I can't think of what it would be.
You see, back when websites were designed around pictures and aligning them to each other....
-
@levicki said in WTF Bites:
Let's see if we can find out:
No, it's reading that info from the S.M.A.R.T. attributes on the drive itself. Nothing will tell you anything more than that,
etymsynology is not at fault for this.My only question is, what did you think the test was going to do? It already says there's pending sectors to reallocate, who cares where? It doesn't actually do the reallocating either. Nor does it "take the drive offline"
-
@levicki said in WTF Bites:
Either don't allow user to run a test while the drive is online and a part of a mirror like it did, or if you already do such a stupid thing then start a test in non-interruptible mode and don't fucking touch the drive -- mark it as dirty and sync it with the other RAID member once the SMART test is done.
This sounds remarkably uninformed, inefficient, and wasteful. "Don't allow user to run a test while the drive is online".
I regularly run both short and long tests on all of my drives while they're being actively used:
Such a stupid @tsaukpaetra, I know. Tell me, do you also take the array offline to do a scrub? Because scrubbing is also something you should be doing somewhat regularly. Oh, I must be stupid for doing that too, eh?
@levicki said in WTF Bites:
I didn't think -- I know very well what SMART extended test does -- it does a full scan of drive surface (hence 8 hours+ for 4 TB), and any bad sectors it finds it tries to read and if it reads them successfully then it reallocates them.
First: Citation needed.
Second: But why would you bother? Grab the spare that I know you have (at risk of me assuming you're stupid and don't have one on hand) and chuck it in. Takes the same time (or LESS!) as a smart test, doesn't faff about with a failing drive, and at worse has the same effect as having taken the drive offline for the test anyways so it's not like you're losing anything in the meantime.And people make fun of me for fucking around with Windows.
@levicki said in WTF Bites:
So I was hoping that it will manage to read and reallocate a pending sector and also that it will tell me if there are more than that one so I know whether I should be in a hurry to replace that particular drive or not.
But again, why? Is your data not important? Apparently not, if "My drive is failing. Meh, let's run a test that takes half a day to see if I want to replace it in the next month or year" is your modus operandi.
@levicki said in WTF Bites:
There is one, but maybe a full scan would reveal more. or maybe even that one would disappear -- I would like to know that.
See above where I mention regularly running said tests in the background automatically and regularly. This should not be a surprise at any time, and knowing that more sectors are being reallocated than the last test should induce no more urgency than the first notice of bad sectors. Dafuq, this isn't a science experiment where you gaze in curiosity at the funny-acting drive. Just fucking replace the damn thing and give the broken one to charity or something!
@levicki said in WTF Bites:
It does if it can manage to read it during the test, otherwise the only way is to try writing something to it to let the drive know you don't care about its contents and then it gets reallocated on write if write fails too. Problem is I don't know which sector is bad because stupid NAS is not telling me its LBA number.
BUT SAID KNOWLEDGE IS TANGENTIAL TO YOU ACTUALLY REPLACING THE DRIVE POSTE HASTE.
See above requesting citation that a long test reallocates the pending sectors, and the admonition that you don't fucking play with your data if it's important to you.
Knowing which LBA is pending/reallocated isn't all that useful regardless, and at best is a tedious way to find which sector to blast a write with so that the RAID can rewrite it for no reason because of course you're replacing that drive as soon as possible, right?
@levicki said in WTF Bites:
By "take offline" I first and foremost meant "out of the mirror for the duration of the test". You know, so it doesn't get interrupted.
So, in other words, you expect the NAS to compromise your data redundancy for what should be regularly-scheduled maintenance. Got it.
@levicki said in WTF Bites:
But also as per smartctl man page you can run the test in different modes, and you can even resume the test if you know where it was interrupted. Synology doesn't do any of those smart things.
It just gives me an equivalent of "Something happened :(".Sure. It would be a nice-to-have, but ultimately pointless.
It would be like implementing a wear-leveler for paper-and-pencil. Once it's used, you toss the thing!
-
@levicki said in WTF Bites:
As for how and when reallocation happens, just google it.
K know when reallocation happens and nothing I've googled suggests a surface read (i.e. what a "long test" is, but on the drive itself) will do as you suggest.
You said it was in a mirror. You didn't say it was btrfs (which is tangential and irrelevant). Doesn't change my statement that
@Tsaukpaetra said in WTF Bites:
So, in other words, you expect the NAS to compromise your data redundancy for what should be regularly-scheduled maintenance. Got it.
@levicki said in WTF Bites:
there is no reason to panic.
You're right, there isn't, because you have a spare ready and willing.
@levicki said in WTF Bites:
Yes I was already planning an upgrade (need more space anyway), but I need some time to find good drives and I am not even sure whether current NAS could even accomodate larger capacity., not to mention that the other drive is perfectly fine.
Completely irrelevant.
Look, I get it, you don't care about your data, that's fine. But blaming the NAS for your bad habits is unfair and disingenuous.
I used to be like you, babying dying drives to get the last electrical bit of usefulness out of them. You should see the "disk massaging" routines I wrote to keep them "healthy".
Then I realized I was spending inordinate amounts of time to life-support equipment I could outright replace for half the cost and a fraction of the time, so every year a family member gets a nice pretty-much-great-condition-except-a-few-bad-or-slow-sectors drive and I get peace of mind.
-
@Tsaukpaetra said in WTF Bites:
You should see the "disk massaging" routines I wrote to keep them "healthy".
The off-by-one thread is
-
@Zerosquare said in WTF Bites:
@Tsaukpaetra said in WTF Bites:
You should see the "disk massaging" routines I wrote to keep them "healthy".
The off-by-one thread is
It sure is! Free Internet pointz! I'll be here all eternity...
-
There's also the common pattern of "curl this shell script directly into your command line as root"
Given that most kinds of packages contain post-inst scripts and that those also run as root, it's not actually any less safe than downloading an installing a
.deb,.rpmor similar package.It's also no different from running an arbitrary
.exeinstaller on Windows as is still plenty common there.
That said, it is a bad idea still, because packages at least have some structure and conventions and keep track of files they installed, though occasionally they still fail to actually do so due to the aforementioned post-install script. Ad-hoc install scripts in contrast tend to do random inconsistent things.
Unfortunately actually sensible installation tools are in short supply. There is snap, but Canonical never did anything really right yet (last time I looked they relied on not so widely used apparmor for security), there is flatpak, which seems geared towards GUI applications (it has better security though), and there kind of is appimage, but really isn't, because it isn't packaged in Debian (nor Ubuntu).
-
@boomzilla said in WTF Bites:
@boomzilla said in WTF Bites:
As encountered today:
fooNumber nvarchar(25)
Government paperwork is full of alphanumeric "numbers".
This column is really numbers, though. Just stored as strings.
And even Unicode strings at that.
How else would you represent a number like ๒๐๑๙?
Which BTW is why parsing stuff you'd like to coerce/stdlib-parse to numeric types out of text using something like
(\d+)is wrong. Depending on the locale it may or may not match stuff like this that's technically "digits" but usual conversion functions are not happy about.
-
-
@boomzilla said in WTF Bites:
@boomzilla said in WTF Bites:
As encountered today:
fooNumber nvarchar(25)Government paperwork is full of alphanumeric "numbers".
This column is really numbers, though. Just stored as strings.
And even Unicode strings at that.
How else would you represent a number like ๒๐๑๙?
Which BTW is why parsing stuff you'd like to coerce/stdlib-parse to numeric types out of text using something like
(\d+)is wrong. Depending on the locale it may or may not match stuff like this that's technically "digits" but usual conversion functions are not happy about.There is a trap joke in here somewhere...
-
Which BTW is why parsing stuff you'd like to coerce/stdlib-parse to numeric types out of text using something like (\d+) is wrong. Depending on the locale it may or may not match stuff like this that's technically "digits" but usual conversion functions are not happy about.
And then you realize that the worst problem in reading numbers is that pesky
,(unless you decide to venture in the non-decimal territory, but everybody uses some form of that these days). Because all the digits from all the writing systems have theNd(Number, decimal digit) category and a unique digit value encoded in their Unicode properties, but the simple ASCII,is used as decimal point in some languages and as thousands separator in others…
-
@levicki So you'll just need to get wireless adapters, you should move with the times!
</sarcasm>
-
@levicki said in WTF Bites:
of the day -- next iPhones for 2020/2021 will be fully wireless (no connectors whatsoever). Such courage!wow.... fully locked into the walled garden too. because there's no USB connection it all has to be over the air.... charging, which is probably something nefarious like standard GI charging, but with some NFC beacon that needs to be present or the charge circuit disconnects so you have to buy first party. only music available via iTunes, only content available via iTV, only television station Fox news (and not the one run by woodland animals reporting woodland news.... the other one)
yeah that's one hell of a walled garden..... and the prisoners of the garden actually like it? that's some serious Stockholm syndrome shit.
-
that's some serious Stockholm syndrome shit.
It's called RDF
And yes, it was invented by Apple (for real this time)
-
Which BTW is why parsing stuff you'd like to coerce/stdlib-parse to numeric types out of text using something like (\d+) is wrong. Depending on the locale it may or may not match stuff like this that's technically "digits" but usual conversion functions are not happy about.
And then you realize that the worst problem in reading numbers is that pesky
,(unless you decide to venture in the non-decimal territory, but everybody uses some form of that these days). Because all the digits from all the writing systems have theNd(Number, decimal digit) category and a unique digit value encoded in their Unicode properties, but the simple ASCII,is used as decimal point in some languages and/or time periods and as thousands separator in others…
-
-
@levicki said in WTF Bites:
of the day -- next iPhones for 2020/2021 will be fully wireless (no connectors whatsoever). Such courage!wow.... fully locked into the walled garden too. because there's no USB connection it all has to be over the air....
Technically, the idea isn't all bad. It's not like you couldn't have a well-documented and open wireless interface just as well as a completely nailed-down USB, and with the proper materials (which nobody uses because they're not shiny enough) you could make some seriously robust devices without flimsy plugs that wear out …
charging, which is probably something nefarious like standard GI charging, but with some NFC beacon that needs to be present or the charge circuit disconnects so you have to buy first party. only music available via iTunes, only content available via iTV, only television station Fox news (and not the one run by woodland animals reporting woodland news.... the other one)
… but Apple being Apple, it's also not like I believed any of those pipe dreams.
-
Butt why
Because fuck you.
Also note the notification period: the corresponding law is dated December 2013, they did fuckall for three and a half years, then notified their customers on May 29 2017 that the change will go live on June 1.
Reality distortion field - Wikipedia
