Return Styles: Pseud0ch, Terminal, Valhalla, NES, Geocities, Blue Moon.

Pages: 1-

C20 is getting 1700 names for floats

Name: Anonymous 2019-10-01 4:57

Name: Anonymous 2019-10-01 6:29

what is the standards committee smoking?

Name: Anonymous 2019-10-01 7:04

>>1
They will be used with underscores like _typename and contained behind _Generic macros.

Name: Anonymous 2019-10-01 7:41

>>3
>Among these identifiers are very short ones, such as dadd, dfma, nanf32, pown, or SNAN, that have a relatively high probability of conflict with user space identifiers.
>Others, such as d64xaddd128x or ufromfpxl are very difficult to parse by the unprepared eye.
>Again others, such as the totalorder functions are “reserved by accident” because the to prefix is generally reserved for external identifiers.
idk about that

Name: Anonymous 2019-10-01 9:25

C is portable assembly created to write operating systems, device drivers, very basic and performance critical tools such as a shell or cat, and rich terminal software for higher-level applications, such as a web browser.

Nobody should care for it otherwise. If you think these types and functions that map to various processors assembly are crazy, don't use C. Use Rust, FIOC, Node.js or anything else. Here are some facts:

1. Nobody cares if your program is 3% slower.
2. If you're a hobbyist, like most people here, nobody cares about your program, period. There's no reason to optimize it. There's no spiritual perfection or nirvana to be achieved.
3. If you're a professional, unless you're doing something huge, development cost and time is way more important than execution speed. OMG! I spent 4 hours making GetCustomerCreditBlock() 3ms faster! Who gives a fucking damn? You're not paid to use C. Write good algorithms and be mindful of scalability (e.g. O(n•log(n)) better than O(n^2)) and don't give a shit about raw speed if it affects readability, stability or dev time.

Name: Anonymous 2019-10-01 9:32

>>5
Time complexity matters a lot less when writing real algorithms than is typically advertised. Its often far more important to optimize for the branch predictor and cache, which means algorithms with low entropy and good locality.
O(nlogn) algorithms tend to do lots of branching, and lose far more to simpler O(n^2) algorithms.

Name: Anonymous 2019-10-01 10:08

>>5,6
all those words your're are writing have nothing to do with the fact that the new standard might break existing old programs

Name: Anonymous 2019-10-01 10:10

>>5
false

Name: Anonymous 2019-10-01 10:58

>>7
I'm neither >>5,6, but you are retarded if you presume AT&T C is anything like C89. Any consequent "C" will never be compatible with any other because there'ren't meant to be forward compatible, because chips are not forward compatible.
>>5 is only correct in that we no longer have portable assembly on chips that emulate their former architectures, and a famous company chose speculative execution against academic and peer professed advice.
Show me a complete compiler up to date covering all OPCODES, then you can post.

Name: Anonymous 2019-10-01 11:20

>>9
if you need to access opcodes directly, just add asm blocks to teh standard.

Name: Anonymous 2019-10-01 11:31

reserved keyword in C22: dubs

Name: Anonymous 2019-10-01 14:05

>>6
For toy sized n. For millions of records?

>>8
Because ___________________.

Name: Anonymous 2019-10-01 16:05

>>10
Thanks for stating the blatant. Still waiting on your compiler.
Programming challenge 98 IIRC.

Name: Anonymous 2019-10-01 16:29

Good thing they finally do something about the current floating point hell and add support for f16, which is widely used in computer graphics and AI.

Name: Anonymous 2019-10-01 20:12

>>12
Depends on the algorithm, but for extremely large data sizes you will run into other even larger costs like reading from secondary storage or requiring map-reduce type solutions.
A lot of industry algorithms use a two step approach. First a divide-and-conquer step to partition the data in manageable blocks, preferably with good locality and entropy. Then, when the problem size is below some threshold, they switch to a dumb algorithm with excellent real world performance because it is tuned for the cache and minimizes branching.

Name: Proud Not-C 2019-10-01 20:42

>>5
C is meaningless niggerbabble, not assembly. Assembly language has semantics and C doesn't. Assembly can use the same mnemonic for different types and C can't. Execution speed is the second most important thing about a program after making sure it works. You give a nigger more development time and they waste it. You give them millions of dollars and they go broke. How does someone make all that money and still be in debt? Those 1700 nigger turds don't "map to" assembly unless you mean the basic job any compiler does. Niggers had 100,000 years of development time and they're still in mud huts. It's time to let the white people program again. C is the nigger. C programs don't work. How do you starve a nigger? Hide his food stamps under his work boots.

Name: Anonymous 2019-10-01 23:08

>>16
C is worse than a nigger, it's a bugchasing nigger with AIDS who enjoys giving other niggers AIDS too.

Name: Anonymous 2019-10-04 14:20

C would be ok if we had:

- Type inference
- Function overloading (types and arity)
- Operator overloading
- Generic types
- References for all values and functions to use instead of pointers
- Lexical scoping and easy let block syntax
- Lambda-expressions returning a reference to a function
- A native dynamic type of string based on dynamic allocation and length-based (not null-terminated)
- A standard library and value syntax for cons cells, linked lists, dynamically allocated lists, sets and dictionaries
- Exceptions, try/catch/finally blocks but no throw Exception Java bullshit on function definitions
- A standard iterator protocol and foreach
- Generators
- Coroutines
- A module system
- An all-new standard types and library that is truly platform-independent and requires no configuration; int is abolished and types without explicit, clear size are abolished
- A standard library with full Unicode semantics and batteries-included modules for JSON, all UNIX semantics and everything else
- while, do, for, if, switch, functions, etc. return the last evalusted value and everything becomes an expression
- Function version of all operators
- A standard library of FP tools
- Optional official GC
- Optional bounds checking
- An improved macro system with better multiline support and gensym
- Perhaps 2 phase compile-time macros that receive AST and can take one or more language constructs after name (e.g. a word, or anything inside parens, brackets or braces)
- Reader macros
- Very good libraries to work with ASTs, and very effective, low-complexity ASTs
- A standard read-compile-dynamic-link-run console
- Compiling flags standardized and can be specified within the source files


I think that's about it.

Name: Proud Not-C 2019-10-04 18:47

>>18
What you're asking for is racist. Niggers are biologically incapable of doing any of that. That's what racism means, recognizing the biological limitations of niggers compared to whites. You can't even expect C to solve the forward reference problem.

Name: Anonymous 2019-10-05 5:31

how offensive

Name: Anonymous 2019-10-05 22:15

>>18
C++ has 80% of these.
Jai has 90% of these.
Rust has 95% of these.

Name: Anonymous 2019-10-05 23:54

>>18
- References for all values and functions to use instead of pointers


You can't iterate an reference. You can do address arithmetic on a reference.

Name: Anonymous 2019-10-06 0:54

>>21
C++ has 800% more things you don't want.
Jai is vaporware.
Farewell sausage!

Name: Anonymous 2019-10-06 23:36

data:image/gif;base64,R0lGODdhMAAwAPAAAAAAAP///ywAAAAAMAAwAAAC8IyPqcvt3wCcDkiLc7C0qwyGHhSWpjQu5yqmCYsapyuvUUlvONmOZtfzgFz ByTB10QgxOR0TqBQejhRNzOfkVJ+5YiUqrXF5Y5lKh/DeuNcP5yLWGsEbtLiOSpa/TPg7JpJHxyendzWTBfX0cxOnKPjgBzi4diinWGdkF8kjdfnycQZXZeYGejmJl ZeGl9i2icVqaNVailT6F5iJ90m6mvuTS4OK05M0vDk0Q4XUtwvKOzrcd3iq9uisF81M1OIcR7lEewwcLp7tuNNkM3uNna3F2JQFo97Vriy/Xl4/f1cf5VWzXyym7PHhhx4dbgYKAAA7

Name: Anonymous 2019-10-07 1:55

>>24
fuck you, I had to throw away my hard drive after this

Name: Proud Not-C 2019-10-07 16:59

>>20
C is offensive by its very existence. This member of the subhuman substandard committee said that carelessly adding 1700 identifiers is bad because of their names. If we rename them to something politically correct, it's totally fine to add 1700 identifiers. This is more TNB. Niggers are obsessed with names. That's where the euphemism treadmill comes from. These 1700 identifiers are a form of welfare created to give C more of the power of white languages. White languages do not need a million different ways to write add or sin because white people know what types are. C never properly understood types. Niggers have the same purchasing power as whites if you give them welfare checks taken from the taxes paid by white people. Hollywood has to cast niggers in TV and movies because niggers get all that money the government stole from white people and they want "a piece of the pie" (see the Jeffersons). Hollywood has to be nice to these niggers in order to get them to work for them which is why there are so many niggerlovers in California. Other niggerlovers start to congregate in the same cities and make everyone into more radical niggerlovers through threats of being called a racist and losing their job. They start replacing white characters with niggers, literally rewriting their scripts in C. It becomes "racist" to even suggest that these characters who were intended as white should stay white.

Name: Anonymous 2019-10-08 1:29

chiggers

Name: Anonymous 2019-10-12 22:31

>>1
This renaming proposal is a heroic effort that developers of long lived code bases will appreciate. It's worrying that the rest of the working group does not appear to be concerned with namespace collisions anymore.

>>4
Among these identifiers are very short ones, such as dadd, dfma, nanf32, pown, or SNAN, that have a relatively high probability of conflict with user space identifiers.
idk about that

I've worked on multiple commercial C code bases that were freestanding for historical reasons and could not be easily ported to a hosted environment because of gratuitous name collisions. Memory gets cheaper (even for embedded systems) but developer time to fix silly compile errors remains expensive.

Name: Anonymous 2019-10-13 10:30

>>18

The result of such changes would not be C anymore. To quote from the C99 standard:

There are many facets of the spirit of C, but the essence is a community sentiment of the underlying principles upon which the C language is based. Some of the facets of the spirit of C can be summarized in phrases like:
Trust the programmer.
Don’t prevent the programmer from doing what needs to be done.
Keep the language small and simple.
Provide only one way to do an operation.
Make it fast, even if it is not guaranteed to be portable.

Name: Anonymous 2019-10-13 10:33

>>29

This is not from the Standard but from the corresponding Rationale: http://www.open-std.org/jtc1/sc22/wg14/www/C99RationaleV5.10.pdf

Name: Anonymous 2019-11-16 20:22

Puny Minds of CIA Cockroaches cannot count to Number 1700. It is not a problem to Libyan Programmers (1700 = 60 + 23 + 990 + 637)

Don't change these.
Name: Email:
Entire Thread Thread List