|
Bongo Bill posted:I � Unicode haha
|
# ? Dec 16, 2021 23:51 |
|
|
# ? May 26, 2024 14:02 |
|
Bongo Bill posted:I � Unicode Volmarias posted:Mods, new thread title, please
|
# ? Dec 16, 2021 23:52 |
|
omeg posted:U+FFFD: "REPLACEMENT CHARACTER", or �. We are switching one of our ancient applications to compile with Java 7 instead of Java 6, and there are a ton of files that no longer compile because they have this stupid replacement character *in comments*.
|
# ? Dec 16, 2021 23:52 |
|
ultrafilter posted:The heaviest book that I've ever seen was a hardcopy of the Unicode standard. Coding Horrors�: A Hardcopy of the Unicode Standard
|
# ? Dec 17, 2021 02:25 |
|
- iMessage wants to make gifs loop forever, even if the file has a finite loop count - their copying method accidentally renders the image instead of just copying and runs outside of sandbox - the rendering library renders based on the actual content, not the claimed filetype - its implementation of a 90s-era BW pdf compression algorithm for scanning text had an overflow bug that lets you access arbitrary offsets from the image being rendered - the image parser has no intentional scripting capabilities, but it does let you do some flexible per pixel bit operations (intended for things like "the pixels for this letter/rectangle are a modified copy of a previous one)" - AND, OR, and XOR are enough to build any logic circuit - the image being rendered into can be used as a very long tape - you can build a computer out of the pixel compression steps that reads system memory as input, uses the image as scratch space, and computes offsets to break ASLR => Receiving a SMS (but not clicking or running anything from it) let the Saudi government spy on some dissidents https://googleprojectzero.blogspot.com/2021/12/a-deep-dive-into-nso-zero-click.html
|
# ? Dec 17, 2021 06:24 |
|
That page also links to "Xerox scanners/photocopiers randomly alter numbers in scanned documents", which I'd never heard of before
|
# ? Dec 17, 2021 10:03 |
|
Hammerite posted:That page also links to "Xerox scanners/photocopiers randomly alter numbers in scanned documents", which I'd never heard of before I fear the day some idiot is going to add machine learning to simple machines. And the only reason they will do that is to have a silver bullet and be able to put "smart" in the name of the product.
|
# ? Dec 17, 2021 14:12 |
|
[deleted, still learning this 'reload before post' thing]
|
# ? Dec 18, 2021 00:39 |
|
Foxfire_ posted:- iMessage wants to make gifs loop forever, even if the file has a finite loop count I was wondering what the XKCD the other day about turing completeness was referencing. This is one hell of a loving bug. Jesus.
|
# ? Dec 20, 2021 21:58 |
|
Blue Footed Booby posted:I was wondering what the XKCD the other day about turing completeness was referencing. I did study an exploit for Gaim? I think? where every byte of the payload needed to pass isalpha(), but since x86 is a terrible old architecture, it's not actually that difficult to build a Turing-complete subset of x86 instructions that pass isalpha(). I think the "exec("/bin/sh")" payload was like 100MB, but there wasn't any limit on the message size.
|
# ? Dec 20, 2021 23:36 |
|
more falafel please posted:I did study an exploit for Gaim? I think? where every byte of the payload needed to pass isalpha(), but since x86 is a terrible old architecture, it's not actually that difficult to build a Turing-complete subset of x86 instructions that pass isalpha(). I think the "exec("/bin/sh")" payload was like 100MB, but there wasn't any limit on the message size. It would be extremely hard to find a subset of x86 instructions that is useful but not Turing complete. Stephen Dolan posted:It is well-known that the x86 instruction set is baroque, overcomplicated, and redundantly redundant. We show just how much fluff it has by demonstrating that it remains Turing-complete when reduced to just one instruction. Paper
|
# ? Dec 21, 2021 03:25 |
|
I am not sure that's actually a single /instruction/ rather than a single assembly mnemonic. As a side note, while CISC can theoretically be better than RISC at code density, I am curious as to how well x86 actually does, given how many of its 1-byte instructions are pretty useless for modern code. It does have a sort of interesting property that bit flips often blow up spectacularly, though. (Obvious comparison is ARM thumb which is optimized for things more modern than 8086). OddObserver fucked around with this message at 04:06 on Dec 21, 2021 |
# ? Dec 21, 2021 03:36 |
|
Zopotantor posted:It would be extremely hard to find a subset of x86 instructions that is useful but not Turing complete. 'h' is pretty useful, because it's push immediate. 'A' is pretty good padding because it's "inc ecx" which is effectively a NOP most of the time.
|
# ? Dec 21, 2021 03:56 |
|
Compiling C to printable x86, to make an executable research paper http://tom7.org/abc/
|
# ? Dec 21, 2021 09:20 |
|
Dylan16807 posted:Compiling C to printable x86, to make an executable research paper http://tom7.org/abc/ in the old good days people would download files from the news://, since is a ascii protocol, these files where codified in 7 bits, so had to be converted with a program into binaries 8 bits but how you boostrap that? you need a executable to turn 7 bits texts in 8 bits binaries. If you don't have that executable, you can download it from the news, and then... ooops so people made a uundecoder.com file in 7 bits that you could download from the news:// and then use it to turn alt.sex.xxx into binaries Tei fucked around with this message at 10:04 on Dec 21, 2021 |
# ? Dec 21, 2021 09:57 |
|
Dylan16807 posted:Compiling C to printable x86, to make an executable research paper http://tom7.org/abc/ This (excellent) paper has my favourite use of a parenthetical question mark: quote:This string is used to test antivirus software, because you can hide this string away inside some file and then see if the antivirus software can successfully find it (?). For some reason this sent me into fits of laughter the first time I read it. I guess I had never considered how nearly useless that test is.
|
# ? Dec 21, 2021 17:50 |
|
Literally just finished building and pushing releases (an arduous task for my area) with log4j 2.16 to find out there was another vulnerability found and addressed over the weekend. Maybe we should just not have logging.
|
# ? Dec 21, 2021 19:49 |
Rubellavator posted:Maybe we should just not have Java. ftfy
|
|
# ? Dec 21, 2021 20:05 |
|
Rubellavator posted:Maybe we should just not ftfy
|
# ? Dec 21, 2021 20:06 |
|
Coding Horrors: Maybe we should just not
|
# ? Dec 21, 2021 22:16 |
|
pokeyman posted:This (excellent) paper has my favourite use of a parenthetical question mark: It has a practical use. It is used by pentesters, when they're testing some business' file upload / cloud storage tooling, to see if they have the most basic of antivirus protections set up.
|
# ? Dec 21, 2021 22:53 |
|
ultrafilter posted:Coding Horrors: Maybe we should just not Coding horrors: Let's not and say we did
|
# ? Dec 22, 2021 01:31 |
|
ultrafilter posted:Coding Horrors:
|
# ? Dec 22, 2021 04:09 |
|
ultrafilter posted:Coding Horrors: !(we should just)
|
# ? Dec 22, 2021 12:04 |
|
Bongo Bill posted:I � Unicode do this one first
|
# ? Dec 29, 2021 19:45 |
|
forums dont support unicode titles
|
# ? Dec 29, 2021 20:02 |
|
taqueso posted:forums dont support unicode titles code:
|
# ? Dec 29, 2021 20:16 |
|
Bongo Bill posted:I � Unicode
|
# ? Dec 29, 2021 20:25 |
|
What's new Unicode, WOOAH WOAAH WWOAAAAH....
|
# ? Dec 29, 2021 20:37 |
|
Had a Unicode coding horror this week. Well the horror was someone with a choice of *nix or Windows for development chose windows but the CI script was choking on commits with Japanese translations of labels because it was in Unicode and the way windows represented it was shock horror not to spec.
|
# ? Dec 29, 2021 22:31 |
|
Which Unicode spec? You might be getting valid UTF-16 but expecting UTF-8. These are not the same things.
|
# ? Dec 30, 2021 00:22 |
|
necrotic posted:You might be getting valid UTF-16 Still counts as a coding horror
|
# ? Dec 30, 2021 00:25 |
|
DoctorTristan posted:Still counts as a coding horror Won't disagree there. But just saying "Unicode" doesn't communicate much.
|
# ? Dec 30, 2021 00:26 |
|
necrotic posted:Won't disagree there. But just saying "Unicode" doesn't communicate much. Maybe they should call it Multicode, then!
|
# ? Dec 30, 2021 00:30 |
|
It had slipped under my radar until pretty recently but Windows 10 actually supports UTF-8 now, you put a special manifest in your executable and the "legacy ASCII" APIs magically start speaking Unicode No more UTF-16 bullshit, what a time to be alive
|
# ? Dec 30, 2021 00:35 |
|
UTF-8 but it was encoded in a Salesforce Metadata xml file for custom labels. Windows is just loving up Japanese characters so now someone not using windows needs to edit those files. Salesforce is also a coding horror but I have stared into the abyss to long so do not mourn for me.
|
# ? Dec 30, 2021 01:32 |
|
I'm for whatever reason reminded of Windows' case insensitivity causing the Perforce client to not commit changes to a file if the casing of the name changed even if the contents of the file were also changed and a bunch of other nonsense, all because the Perforce server was running on linux with case sensitivity. Looking back, wow what a massively oversold pile of poo poo Perforce turned out to be and I regret ever suggesting it.
|
# ? Dec 30, 2021 01:58 |
|
Perforce is great! It's wonderful! It's great! I love using it! Love to have utter messes integrating because someone renamed a file!
|
# ? Dec 30, 2021 02:26 |
|
I used SVN and CVS before GIT was a thing and in my mind that is remembered as the dark times.
|
# ? Dec 30, 2021 02:36 |
|
|
# ? May 26, 2024 14:02 |
|
Best part of my recent job change is I no longer have to use Perforce
|
# ? Dec 30, 2021 02:38 |