| |
| |
| |
|
Comments:
<donfede> hello folk. i am working on creating a pam module . in that module i need to use both C and C++ ... C for the pam connections, and C++ for the project it's part of (using QT and several other C++ libraries)
<donfede> i am running into problems linking the two ... i have wrapped the pam C functions in the module source code within ' extern "C" { } ' blocks ...
<donfede> and as long as I don't link any libraries (-lfoo on the g++ command line), it seems to work fine ...
<donfede> but once I add a library (-lqui or -lrd) i get PAM unable to dlopen(/lib/security/pam_rd.so)
<donfede> PAM [dlerror: /lib/security/pam_rd.so: invalid ELF header]
<donfede> so the solution to my problem was pretty much to compile the module using g++ ... my error was :'( using the "libtool bash shell script" instead of the actual library in .libs/
<buck68> I'm having a crazy problem with casting ints to doubles in C... any reason why casting an int to double should produce a NaN?
<pfeilspitze> how are you casting?
<buck68> pfeilspitze: i've tried all sorts of variations... but the original was jsut fsecs = (double)tv->tv_sec. fsecs is a double. tv is struct timeval *
<buck68> i don't trust "isnan()" so much right now... i think there is some kind of optimization bug.
<lotek> I get a rather strange error when attempting to configure gcc 4.1.1 from source contained in gcc-4.1.1.tar.bz2.
<lotek> configure: error: can not find install-sh or install.sh in . ./.. ./../..
<ferguscan> Anyone have experience with mudflap and threaded programs? It's not behaving for me (false negatives), and I'm wondering if it's a Debian-specific problem.
<ferguscan> Could someone run a small test app on a non-Debian machine for me?
<vrt> guys, how do i compile gcc on a box where the gcc package is not installed?
<tdalman> hello
<ferguscan> Anyone here have experience with mudflap? It's not behaving itself under threaded environments (4.1.1)
<tehxeer> are there 64bit ports floating around out there? (windows)
<tehxeer> i tried djgpp, cant even execure the image
<tehxeer> execute*
<Zordan> djgpp is the DOS port :)
<tehxeer> is there a win64 port =)
<tehxeer> djgpp was working fine for me when i was win32
<Zordan> yeah 32 bit windows can run dos binaries
<Zordan> mingw is for win32 ... not sure if there is a 64 bit version yet
<Zordan> but it doesn't look like there is
<Zordan> according to gcc's documentation only linux and free/net BSD are currently supported for x86_64
<tehxeer> seems i fixed the gcc issue, now i just need to find proper nasm and ld images
<tehxeer> those are not bundled with mingw, are they?
<Zordan> there is a bundle which includes gcc/binutils/win32headers and whatnot
<Zordan> this one: http://prdownloads.sf.net/mingw/MinGW-3.1.0-1.exe?download
<Zordan> and additionaly this one if you like bash and autotools: http://prdownloads.sf.net/mingw/MSYS-1.0.10.exe?download
<GyrosGeier> hi
<GyrosGeier> does anyone understand how -fno-exceptions works?
<GyrosGeier> I have a piece of code with a function try block around main()
<GyrosGeier> when I use -fno-exceptions, I get an error message 'expected init-declarator before "if"'
<GyrosGeier> the entire program does not contain the letters "i", followed by "f"
<Zordan> with -fno-exceptions try and catch blocks are substituted by if blocks
<Zordan> try is replaced by if(true) and catch is replaced by if(false)
<alfadir> looking into a larger project doing source to source tranlations. have printed gimple trees and so, and read about pretty-printer. anyone have any more direct pointers where to look or who to talk to ?
<GyrosGeier> Zordan, ah, thanks
<GyrosGeier> Zordan, that is exactly what I need
<Zordan> yw
<GyrosGeier> hm
<GyrosGeier> it appears that it does the same for function try blocks
<Zordan> take a look at the preprocessed code
<Zordan> (substitution is done by the preprocessor)
<GyrosGeier> ah
<GyrosGeier> the preprocessor wouldn't know whether it's a normal try/catch or a function try block
<GyrosGeier> because "if" before '{' does not work.
<GyrosGeier> hm
<mElo97> (newbie) what's a "function try block"?
<GyrosGeier> even with -fno-exceptions I still have a stinking big pile of code
<GyrosGeier> in fact, I only gained 400 bytes
<Zordan> got rid of rtti?
<GyrosGeier> mElo97, you put the "try" before the opening block of the function
<GyrosGeier> Zordan, yes
<GyrosGeier> Zordan, at least I tried
<GyrosGeier> Zordan, I still have lots of typeinfo structures in the binary
<Zordan> -fno-rtti?
<GyrosGeier> Zordan, yep
<Zordan> hm -Os and upx? :)
<GyrosGeier> -Os gains another 400 bytes
<GyrosGeier> upx does not make sense since it doesn't handle my architecture
<Zordan> which is?
<GyrosGeier> ARM bigendian uClinux
<GyrosGeier> oh, and no ELF binaries either
<GyrosGeier> ELF does not deal with MMUless arches really
<Zordan> well that is problematic then
<GyrosGeier> the problem is that it's a program that prints the time.
<GyrosGeier> It's 400 KB.
<Zordan> heh
<GyrosGeier> (statically linked)
<GyrosGeier> and fully stripped
<GyrosGeier> because the executable format (bLFT) consists of just a 64 byte header in front of the data, and 4 bytes per relocation
<mElo97> GyrosGeier, thx I didn't know about it. Found much info on the web... ;-
<GyrosGeier> mElo97, it helps you save a layer of indentation at worst, and IIRC you can also catch exceptions in parameter handling or so this way
<GyrosGeier> oh wait
<GyrosGeier> no, parameters are before
<GyrosGeier> because you can access them from the catch block
<mElo97> I'm just reading that, and was seeking for the usefulness of a func-try block: Indeed as you said "bad parameter handling" could be implemented this way as a throw() from inside the function body can be caught by the catch() "~counterpart". thx
<mElo97> euh
<mElo97> I might have confused a few things, please disregard..
<GyrosGeier> mElo97, yes; parameter handling is, however, the caller's duty
<GyrosGeier> mElo97, (parameter p***ing can throw as well if you p*** an object by value)
<mElo97> GyrosGeier, oh :- you mean if I p*** an object by value, and the constructor of this object throw() something! I never thought about this case.. ok
<GyrosGeier> mElo97, if you want to have really interesting and funny effects, p*** a std::auto_ptr by value.
<mdmkolb1> Is this a GCC bug? http://cpp.sourceforge.net/?show=17184
<unlink> make segfaults when i set --prefix=/
<unlink> gcc/xgcc specifically
<unlink> make bootstrap, sorry
<taker12> How can I use win32 dlls when writing a program in the Linux environment?
<drewski> does anyone know where I can find the docs on how gcc 3.4.4 mangles names?
<Zordan> why?
<Zordan> or rather what for?
<drewski> I am seeing how difficult it would be to make freepascal able to link to c++ objects in shared object files
<drewski> this is most likely beyond my abilities but it doesn't hurt to look :)
<Zordan> hm I don't know anything about freepascal or its ABI :/
<drewski> apparently some time ago there was an effort to accomplish this but since then the name mangling for g++ has changed so I'm trying to see if I can update it
<drewski> it = freepascal
<mdmkolbe|work> drewski: it might be easier to write freepascal binding set for SWIG (http://www.swig.org/). just a thought
<Zordan> drewski: http://www.codesourcery.com/cxx-abi/abi.html#mangling
<Zordan> that's for gcc4 at least ... not sure about 3.4.4
<drewski> Zordan: thank you ver much!
Return to
#gcc
or
Go to some related
logs:
#fedora
#linuxhelp
tablecontentparser ubuntu
grub remove the border
#css
#perl
#physics
php strtoproper
linux wipe ram
#css
|
|
