Dirty Old Joseph the Unix Hacker

DIS ONES FOR ALL DA RETOIDS IN ##c@irc.freenode.net AND ALL DA RETOIDS IN comp.lang.c.

Still using his mainframe, still hanging onto scraps.

Still reinventing libc for all of his apps.

The chortling, snorting, farting old coote

has been shitting on the Internet for longer than moot.

Where you were lamda? I missed you ;-;

I WAS IN HAWAII.

I HAD IT UP TO HERE *GESTURES TO MY EYE LEVEL* WIT DA FUCKIN STACK BOYS, DA FUCKIN MATH BOYS, N DA FUCKIN "KAWAII SAFARII" JAPANIMATION RETOIDS.

bitch, im a robot

Hey, I'm willing to read and follow the standard, but leave my fucking animes out of this.

Lambda, what would you change in C if you could?
I would remove the signal shit, the static buffer functions, the non-thread safe functions and the stupid int, short, long, etc... and just give to the user the option to define a n-bit type

I would remove the signal shit
ITS NOT LIKE ITS ALL DAT USEFUL ANYWAY

the static buffer functions
WAT DA FUK R DEY?

non-thread safe functions
WELL DONT BLAME ME FOR strtok. DEY DIDNT LIKE DA ALTERNATIVE I PROPOSED COS IT DIDNT COMPLY WITH EXISTING CONVENTIONS.

just give to the user the option to define a n-bit type
I WUDNT GO DAT FAR. DATS 4 DAM SURE. I WUD REQUIRE DA IMPLEMENTATION 2 DEFINE AN UNSIGNED, 8-BIT UNPADDED TYPE, WHICH CUD BE WRITTEN TO BINARY-MAPPED STREAMS. DATS ALL U NEED TO PORTABLY REPRESENT ALL DA COMMONLY USED FLOATING POINT AND SIGNED INTEGER REPRESENTATIONS, ALONG WITH ALL DA DIFFERENT BYTE ORDERS USED 2 STORE DOSE REPRESENTATIONS. ASIDE FROM DAT, OBJECT REPRESENTATION CUD BE ENTIRELY UP TO DA IMPLEMENTATION. EITHER DA STANDARD C LIBRARY (OR ANOTHER LIBRARY) CUD DEN PROVIDE FUNCTIONS, WITH A PORTABLE IMPLEMENTATION, DAT READS AND WRITES TO COMMON FORMATS 4 EASE OF USE.

unsigned char WORKS FOR DIS PURPOSE ANYWAY. ONLY DIRTY OLD JOSEPH DA UNIX HACKER USES SHIT FOR WHICH CHAR_BIT AINT 8.

ANOTHER MAIN PROBLEM, DAT TIES INTO DIS, IS DA SHITTY <stdio.h> INTERFACE. ALL DAT FILE REPOSITIONING CRAP DOES MORE HARM THAN GOOD N SHUD BE REMOVED. SOMETHIN LIKE A "stream ready?" FUNCTION (SIMILAR TO DA char-ready? PROCEDURE IN SCHEME) OR A select() OPTION WUD HAVE BEEN A GOOD IDEA. SHUD BE WELL DEFINED ACCORDING TO BUFFERING OPTIONS, LIKE IF U HAVE LINE BUFFERING SET AND DA STREAM'S READY, U CAN READ AN ENTIRE FUCKING LINE (WITH A FUNCTION LIKE fgets) WITHOUT IT BLOCKING.


IF DERS ANYTHING I HATE ITS DA FUKIN LONG unsigned/signed TYPE NAMES. I LIKE DA WAY PLAN9 WENT ABOUT TYPEDEFING SHIT LIKE "unsigned long" AND "signed char" AS "ulong" AND "schar". EASIER 2 READ AND 2 TYPE.

>>7
I WUD REQUIRE DA IMPLEMENTATION 2 DEFINE AN UNSIGNED, 8-BIT UNPADDED TYPE
YOU MENA uint8_t?

where can i read the standard?

>>9
ANSI X3.159-1989, ISO/IEC 9899:1990, ISO/IEC 9899:199409, ISO/IEC 9899:1999 or ISO/IEC 9899:2011?

>>8
YAINT RED DA FUCKIN STANDARD. uint8_t AINT REQUIRED BY DA FUCKIN C IMPLEMENTATION.

>>9
JUST GO 2 ONE OF DA COMMITEEE SESSIONS N STEEL A COPY FROM WUN OF DOSE RETOIDS WHO REPRESENT MICROSOFT.

>>11
How would you have an ``UNSIGNED, 8-BIT UNPADDED TYPE'' if CHAR_BIT is 9?
Ya fucking stackboy

>>12
CHAR_BIT CUDNT BE 9 IN DAT SCENARIO. U WUD HAVE TO EXPRESS SIZES IN TERMS OF DA 8-BIT TYPE AND SINCE 9 AINT A MULTIPLE OF 8, BYEBYE DIRTY OLD JOSEPH DA UNIX HACKER.

ESSENTIALLY, IF DA STANDARD JUST REQUIRED CHAR_BIT == 8, unsigned char WUD MEET DA FUCKING CRITERION I DESCRIBED.

NOW GET DA FUK OUTTA MY THRED YA FUCKIN RETOID

>>13
I HATE YOU! YOU ARE JUST LIKE THE FUCKIN ``R..'' FROM STACK OVERGROW
YOU WANT TO MAKE ME NOT TO ABLE TO PROGRAM IN C ON MY PDP-10? IS THIS WHAT YOU WANT? YA FUCKIN STACKBOY

>>14

Still using his mainframe, still hanging onto scraps.

Still reinventing libc for all of his apps.

The chortling, snorting, farting old coote

has been shitting on the Internet for longer than moot.

Dirty old Joseph the Unix hacker.

Dirty old Joseph the Unix hacker.

>>15
unix hacker
Shalom!
Posix forces CHAR_BIT == 8

>>7
Why didn't you stab the guy who did <errno.h> in the eyes? You should've stab'd him good.

>>17
atleast errno is thread-safe in C11 and in POSIX ^^

>>11
uint8_t AINT REQUIRED BY DA FUCKIN C IMPLEMENTATION
It is required in the case CHAR_BIT == 8
Read ISO/IEC 9899:2011 7.20.1.1.3
``These types are optional. However, if an implementation provides integer types with widths of 8, 16, 32, or 64 bits, no padding bits, and (for the signed types) that have a two’s complement representation, it shall define the corresponding typedef names.''

>>18
Only because of hacks. It pisses me off that implementations are permitted to magically define thread local variables with global scope while the programmer is forced to use pthread getters and setters.

C11 finally fixed this by adding thread_local. Six years isn't a long time when you're talking about DA STANDARD.

>>20
in posix errno is not a variable but a macro to a function that uses pthread things

https://stackoverflow.com/questions/12945486/is-errno-legal-c

>>19
AND CHAR_BIT == 8 AINT REQUIRED SO DATS Y I WUD MAKE IT SO DAT IT WERE

>>17,20
IF U WANT CUTE THREDS UR BETTER OFF USING A CUTE LANGUAGE LIKE GO. DOSE HYENAS AT DA ISO WANT C TO BE A BLOATED HUNK OF UNUSEABLE SHIT, SO DAT'S Y DEY'VE BEEN FUCKIN IT UP WITH THREDS, N DAT'S Y DEANIS RICKY DID HIMSELF IN. DER'S SERIOUSLY NO NEED FOR C TO SUPPORT THREDS. U HAVE FUCKIN pthread IF U *REALLY* NEED DAT SHIT, BUT TRADITIONALLY PEOPLE WHO WERE WRITIN DAEMONS HAVE USED EITHER fork() OR select(), AND DAT'S WORKED REAL FUCKIN WELL WITH C PROGRAMS.

DA PEOPLE WHO WANT THREDS R MOSTLY DA SAME RETOIDS WHO DESIGN HUGE USERSPACE PROGRAMS DAT FUNCTION LIKE MINI OPERATING SYSTEMS. DESE RETOIDS SHUD BE POINTED AT REMEDIAL PROGRAMMING COURSES, NOT ENCOURAGED BY STUPID SHIT LIKE DA <threads.h> INTERFACE.

C LIBRARY IMPLEMENTERS JUST CANT BE FUCKED IMPLEMENTING <threads.h> COS 1) DA IDEA OF PUTTING THREDS IN C IS A FUCKIN JOKE, N 2) <threads.h> IS A FUCKIN JOKE IN GENERAL. PUTTIN THREDS IN C++ IS FITTING COS C++ IS A BLOATED HUNK OF SHIT, SO DA MORE USELESS FEATURES DEY CHUCK INTO DA C++ STANDARD DA BETTER. BUT FOR C, <threads.h> IS A FUCKING DISGRACE.

WEN U THINK ABOUT ALL DA STUPID LITTLE CHILDREN WHO R IN CHARGE OF CHANGING DA C STANDARD, IT'S NO FUCKING WONDER DAT DEANIS RICKY DID HIMSELF IN. DESE HYENAS ARE JUST TRYING TO DESTROY DA LANGUAGE DAT DEANIS RICKY LOVED IN DA SUBTLEST WAY POSSIBLE. HOW COME PEOPLE WHOSE OPINIONS ACTUALLY MATTER, LIKE US FUCKERS WHO HUNG OUT AT BELL LABS AT DA TIME, AINT ON DA COMMITTEE ANYMORE? N HOW COME WE HAVE A BUNCH OF MICROSOFT'S DRONES -- HUNGARIAN-NOTATION-WIELDING BARBARIANS -- CALLIN DA SHOTS INSTED?

IF ANYTHING, DA STANDARD'S JUST GOTTEN SHITTIER AND SHITTIER AFTER EACH MAJOR PUBLICATION.

>>23
AND CHAR_BIT == 8 AINT REQUIRED SO DATS Y I WUD MAKE IT SO DAT IT WERE
the only thing by forcing uint8_t (or something like this) is to make CHAR_BIT == 8 forced

The only thing that would happened is to make C less portable by makeing it not to work on CHAR_BIT > 8 systems

U CAN STILL IMPLEMENT 8-BIT ADDRESSING ON SYSTEMS WITH CHAR_BIT > 8.

DA DIFFERENCE IS DAT DA BURDEN IS PLACED ON DA C IMPLEMENTER, NOT DA C PROGRAMMER (WHICH IS WHERE DA BURDEN IS PLACED BY DA STANDARD).

>>25
U CAN STILL IMPLEMENT 8-BIT ADDRESSING ON SYSTEMS WITH CHAR_BIT > 8
By making a virtual macine or some shit. Still, why not both?
they can still have 8 bit bytes and the emulation thing

>>21
Specifically, a macro that calls a function in the thread implementation returningint * and dereferences the result. I'd call that a hack.

>>23
<thread.h> is mostly pointless. On the other hand, something like <stdatomic.h> has really been needed for years.

>>26
making a virtual macine or some shit
As I recall, the C implementation for the PDP-10 (yes, there was one) had to implement its own byte addressing. Otherwise just setting CHAR_BIT to 36 would have wasted ridiculous amounts of memory for string operations while creating ABI issues with other code that expected packed strings.

>>27
Otherwise just setting CHAR_BIT to 36
YOU MENA 9

>>27
>le pedophile sage

>>27
<stdatomic.h>
cancer

>>30 just don't start cooking meth

>>27
No I mena 36. The PDP-10 could only address memory one word at a time. Breaking apart a word into smaller bytes required special instructions.  C implementations on most other systems assume that each char can be identified by a unique machine pointer value; if you tried to do that on a PDP-10 your chars would be 36 bits.

>>30
Would you prefer to implement your own memory barriers using platform specific inline assembly?  If you're using C to replace assembly as a systems language it needs to have atomics.

wow a programming thread on prog
thanks LA

>>32
>le pedophile sage

>>33
>le pedophile sage

>>32
Atomic variables are useless and unportable. They promote the notation of global variables and sharing memory between threads
they are also useless if you don't use threads

PDP-10
Shalom!

>>35
The whole point of having stdatomic.h is to make atomics more portable.  It's true that user programs don't need atomics, but if you're writing an operating system you must have them.  Even single threaded multiprocessing depends on atomics for synchronization of kernel tasks.

>>37
if you're writing an operating system you must have them
not him but why?

Even single threaded multiprocessing depends on atomics for synchronization of kernel tasks
how?

>>38
You need atomics to implement spinlocks and mutexes.  Even the simplest single threaded user program will rely on multiple kernel threads to perform I/O on its behalf, and those kernel threads need to use synchronization primitives.  Hardware is the ultimate form of shared data.

>>39
fug