Survey of Commonly Available C System Header Files | Hacker News Reader

Survey of Commonly Available C System Header Files

hacks.owlfolio.org

58 pointsMattJ10020 comments

This could be really useful for me but the presentation breaks (at least on Chrome). By the time you're halfway down the list it's hard to tell which header's row belongs to which compiler's column.

It also renders extremely bad on Opera/Win7x64 (Table Headers totally cut off) and is sluggish to navigate. Looking at the source I got why that is.

Really good idea, but the presentation needs a do-over.

Yeah, this is definitely a header problem with the header on the table about header problems.

Scroll down works in FF (20.0) but not in Chrome (26.0...) on a Mac.

Does not work in FF 20.0 on Ubuntu. (The header shrinks so that the labels no longer line up with the columns.)

I'm surprised by the small number of "unusably buggy" headers. Maybe it's due most platforms being reasonably recent.

Otherwise I'd say bringing msvc to c compiler comparison is like bringing knife to gunfight.

MSVC isn't a C compiler. It's a compiler for The Language Formerly Known As C.

MSVC does not exist.

The only product Microsoft sells is called "Microsoft Visual C++".

yeah, and GCC is gnu compiler collection. that doesn't stop everybody calling microsoft visual c++ msvc and gnu compiler collection gcc.

The issue is that GCC was once upon a time only "Gnu C Compiler", whereas Visual C++ was always named like that.

It would be nice to see a POSIX.1-2008 section too -- many of the headers listed as "POSIX.1-2001 optional" are required in POSIX.1-2008. (e.g., <aio.h>, <cpio.h>, <dlfcn.h>, <iconv.h>, <langinfo.h>... and <pthread.h>, which is probably the most commonly used 2001-optional/2008-required header).

Useful reference, and whilst controversial - i'd like to have seen ssl.h/openssl.h aswell. But anyway, thanks. :)

It'd be great to see a metric of "completeness". For instance, while <semaphore.h> declares support for unnamed semaphores on OS X, the platform does not have them.

Nice, although I find a few issues:

- A few commercial compilers are missing.

- How much of POSIX does SUA in Windows support?

Hello.

This is really nice, I just have one question, ( maybe I have overlooked it), where is the test-suite for gathering the data?´, as to which headerfiles are workable or not? Is there a configure script there? I mean, I am on Snow Leopard, and I have no trouble using unistd.h which is marked as flawed on the graph.

See the second paragraph, where it states..

https://github.com/zackw/header-survey/

Hello.

All I see that is that it is a survey, I wonder what kind of tests that are used to draw the conclusions. I also see for instances that the regexp.h file is absent for OS X Lion, which is reworked to work wide wide characters. But, the header files may fail in ways I don't know about, that is why I ask for the tests that makes up for the results.

If you'll look at the code in the link posted previously, it appears that it is testing for the presence of the headers rather than functionality inside of them.

> I am on Snow Leopard, and I have no trouble using unistd.h which is marked as flawed on the graph.

The whole row for `unistd.h` is green (except for MSVC, of course) -- why would you think it's flawed on OS X?

Does Cygwin GCC include the same set of posix headers as Unix GCC?

Reply on news.ycombinator.com