libpng-1.5.6rc01 is available

[Nicolas Limare-3]

On Fri, Oct 28, 2011 at 10:51:55AM +0900, Nicolas Limare wrote:
> > Did this compiler work before, what version of libpng?
>
> I don't know, it's the first time I try it on libpng. I will try on
> previous versions.

1.5.6rc01 : ld prints warnings, check succeeds
1.5.0     : ld prints warnings, check succeeds
1.4.8     : ld prints no warning, check succeeds
1.4.0     : ld prints no warning, check succeeds

This is with LD_LIBRARY_PATH correctly set, sorry for the false alarm
I never had this problem with my code.

--
Nicolas LIMARE
http://nicolas.limare.net/                         pgp:0xFA423F4F


------------------------------------------------------------------------------
The demand for IT networking professionals continues to grow, and the
demand for specialized networking skills is growing even more rapidly.
Take a complimentary Learning@Cisco Self-Assessment and learn
about Cisco certifications, training, and career opportunities.
http://p.sf.net/sfu/cisco-dev2dev
_______________________________________________
png-mng-implement mailing list
png-mng-implement@...
https://lists.sourceforge.net/lists/listinfo/png-mng-implement

[John Bowler-3]

From: Nicolas Limare [mailto:nicolas.limare@...]
>1.5.6rc01 : ld prints warnings, check succeeds
>1.5.0     : ld prints warnings, check succeeds
>1.4.8     : ld prints no warning, check succeeds
>1.4.0     : ld prints no warning, check succeeds

If 'check' passes it is fine.  The warnings relate to the version script, which apparently has been messed up because the icc preprocessor does something to '@' signs.  In fact ignoring the characters ends up being harmless, although there would be a problem if the version script requires symbols to be modified from the original C function names.

It's still curious that it works at all, because I was sure pnglibconf.h required the same behavior of the preprocessor, but if it works, it works; supporting icc on !Windows (i.e. other than the existing scripts/makefile.intel support) is new, so fixes should probably wait for 1.5.7.

John Bowler <jbowler@...>



------------------------------------------------------------------------------
The demand for IT networking professionals continues to grow, and the
demand for specialized networking skills is growing even more rapidly.
Take a complimentary Learning@Cisco Self-Assessment and learn
about Cisco certifications, training, and career opportunities.
http://p.sf.net/sfu/cisco-dev2dev
_______________________________________________
png-mng-implement mailing list
png-mng-implement@...
https://lists.sourceforge.net/lists/listinfo/png-mng-implement

[John Bowler-3]

From: Nicolas Limare [mailto:nicolas.limare@...]
>This solved the Clang compiler warnings.
>This compiler defines this identifier:
>#define __clang__ 1

The git head now contains a change that removes the need to identify the compiler.

I'm still mystified that a ?: expression in an initializer has both sides examined, while the ANSI-C (1989) document says that only one side of ?: is evaluated.  The 1989 standard even includes this example of a valid initialization, albeit in a footnote:

static int i = 2 || 1 / 0;

("The expression is a valid integer constant expression with value one.")

John Bowler <jbowler@...>



------------------------------------------------------------------------------
The demand for IT networking professionals continues to grow, and the
demand for specialized networking skills is growing even more rapidly.
Take a complimentary Learning@Cisco Self-Assessment and learn
about Cisco certifications, training, and career opportunities.
http://p.sf.net/sfu/cisco-dev2dev
_______________________________________________
png-mng-implement mailing list
png-mng-implement@...
https://lists.sourceforge.net/lists/listinfo/png-mng-implement

[John Bowler-3]

From: Nicolas Limare [mailto:nicolas.limare@...]
>1.5.6rc01 : ld prints warnings, check succeeds
>1.5.0     : ld prints warnings, check succeeds
>1.4.8     : ld prints no warning, check succeeds
>1.4.0     : ld prints no warning, check succeeds

1.5 started using @@@ as a safe token to indicate concatenation.

This is fine with the Intel compiler just so long as the @@@ does not occur in a preprocessor line that it actually processes; I guess that means so long as it doesn’t occur in a line that the compiler runs the preprocessor tokenizer on to produce pp-tokens.  However if @@@ is used in such a context, specifically in the body of a macro that gets expanded, for some unknown reason the compiler puts a space between adjacent @ signs.  It doesn't put a space anywhere else, it doesn't put a space between pp-tokens (for example 1+2) - something that it is allowed to do.  It doesn't put a space before the @ sign or after, only between two adjacent tokens.

For the moment there are two work rounds:

1) Ignore it: it only changes libpng.vers and the compiler happily ignores the @ signs (with 693 warnings.)
2) When running configure use this:

CC=icc CPP=/lib/cpp ./configure

(Assuming you have a /lib/cpp, another possibility if not is to use "gcc -E", or, indeed -E on almost any compiler.)

I'll probably change it  in 1.5.7 so that everything uses completely valid ANSI-C tokens, something like PNG__CONCAT__

John Bowler <jbowler@...>



------------------------------------------------------------------------------
The demand for IT networking professionals continues to grow, and the
demand for specialized networking skills is growing even more rapidly.
Take a complimentary Learning@Cisco Self-Assessment and learn
about Cisco certifications, training, and career opportunities.
http://p.sf.net/sfu/cisco-dev2dev
_______________________________________________
png-mng-implement mailing list
png-mng-implement@...
https://lists.sourceforge.net/lists/listinfo/png-mng-implement