Impulse convolver/guitarix

[brummer]

run ./waf configure --cxxflags="-O3 -Wall -march=native" again,
dont use the -V4.3

That's allready the new source now in /trunk, it have replace the old one, the structure is somewhat different from the old style, it will be wise to make a new fresh checkout to a empty directory.
You can savly use your old prestets with the new guitarix

so best will be you try first to build it without involce a special compiler, let us know the result so that we can re-akt if somthing go's wrong.

brummer

[funkmuscle]

./waf build
[ 1/31] cxx: ladspa/crybaby.cpp -> build/default/ladspa/crybaby_1.o
[ 2/31] cxx: ladspa/distortion.cpp -> build/default/ladspa/distortion_2.o
[ 3/31] cxx: ladspa/echo.cpp -> build/default/ladspa/echo_3.o
[ 4/31] cxx: ladspa/freeverb.cpp -> build/default/ladspa/freeverb_4.o
[ 5/31] cxx: ladspa/impulseresponse.cpp -> build/default/ladspa/impulseresponse_5.o
[ 6/31] cxx: ladspa/monoamp.cpp -> build/default/ladspa/monoamp_6.o
[ 7/31] cxx: ladspa/monocompressor.cpp -> build/default/ladspa/monocompressor_7.o
[ 8/31] cxx: ladspa/guitarix-ladspa.cpp -> build/default/ladspa/guitarix-ladspa_8.o
[ 9/31] cxx: src/gx_globals.cpp -> build/default/src/gx_globals_1.o
In file included from ../src/headers/guitarix.h:41,
from ../src/gx_globals.cpp:40:
../src/headers/gx_engine.h: In function ‘void* gx_engine::aligned_calloc(size_t, size_t)’:
../src/headers/gx_engine.h:46: error: ‘calloc’ was not declared in this scope
../src/gx_globals.cpp: At global scope:
../src/gx_globals.cpp:198: error: ‘getenv’ was not declared in this scope
Build failed
-> task failed (err #1):
{task: cxx gx_globals.cpp -> gx_globals_1.o}

Fresh checkout also....

[brummer]

okay, update, try again.

gcc 4.4 have change a bit in relation to gcc 4.3 , it's a additional header what gcc 4,4 need. Have include it now. May there are more missing headers, thats posible, let us know

brummer

[funkmuscle]

looks like more are missing:

./waf build
[ 9/31] cxx: src/gx_globals.cpp -> build/default/src/gx_globals_1.o
[10/31] cxx: src/gx_child_process.cpp -> build/default/src/gx_child_process_1.o
In file included from ../src/headers/guitarix.h:41,
from ../src/gx_child_process.cpp:41:
../src/headers/gx_engine.h: In function ‘void* gx_engine::aligned_calloc(size_t, size_t)’:
../src/headers/gx_engine.h:46: error: ‘calloc’ was not declared in this scope
../src/gx_child_process.cpp: In function ‘pid_t gx_child_process::gx_find_child_pid(const char*)’:
../src/gx_child_process.cpp:217: error: ‘atoi’ was not declared in this scope
../src/gx_child_process.cpp: In function ‘void gx_child_process::gx_show_jack_capture_gui(GtkWidget*, void*)’:
../src/gx_child_process.cpp:228: error: ‘system’ was not declared in this scope
../src/gx_child_process.cpp: In function ‘bool gx_child_process::gx_capture_command(int, std::string&)’:
../src/gx_child_process.cpp:267: error: ‘getenv’ was not declared in this scope
../src/gx_child_process.cpp: In function ‘void gx_child_process::gx_start_stop_jconv(GtkWidget*, void*)’:
../src/gx_child_process.cpp:478: error: ‘getenv’ was not declared in this scope
Build failed
-> task failed (err #1):
{task: cxx gx_child_process.cpp -> gx_child_process_1.o}

[brummer]

yes, indeed, so have update again.

one more run ?

brummer

[funkmuscle]

updated and still:

./waf build
[ 9/31] cxx: src/gx_globals.cpp -> build/default/src/gx_globals_1.o
[10/31] cxx: src/gx_child_process.cpp -> build/default/src/gx_child_process_1.o
[11/31] cxx: src/gx_engine.cpp -> build/default/src/gx_engine_1.o
[12/31] cxx: src/gx_gui.cpp -> build/default/src/gx_gui_1.o
In file included from ../src/headers/guitarix.h:41,
from ../src/gx_gui.cpp:41:
../src/headers/gx_engine.h: In function ‘void* gx_engine::aligned_calloc(size_t, size_t)’:
../src/headers/gx_engine.h:46: error: ‘calloc’ was not declared in this scope
Build failed
-> task failed (err #1):
{task: cxx gx_gui.cpp -> gx_gui_1.o}

but will keep trying...

[brummer]

well, so let's go one more try now

[funkmuscle]

./waf build
[12/31] cxx: src/gx_gui.cpp -> build/default/src/gx_gui_1.o
[13/31] cxx: src/gx_jack.cpp -> build/default/src/gx_jack_1.o
[14/31] cxx: src/gx_jconv_settings.cpp -> build/default/src/gx_jconv_settings_1.o
In file included from ../src/headers/guitarix.h:41,
from ../src/gx_jconv_settings.cpp:46:
../src/headers/gx_engine.h: In function ‘void* gx_engine::aligned_calloc(size_t, size_t)’:
../src/headers/gx_engine.h:46: error: ‘calloc’ was not declared in this scope
../src/gx_jconv_settings.cpp: In constructor ‘gx_jconv::GxJConvSettings::GxJConvSettings()’:
../src/gx_jconv_settings.cpp:64: error: ‘getenv’ was not declared in this scope
../src/gx_jconv_settings.cpp: In function ‘void gx_jconv::gx_acquire_jconv_value(GtkWidget*, void*)’:
../src/gx_jconv_settings.cpp:467: error: ‘atoi’ was not declared in this scope
Build failed
-> task failed (err #1):
{task: cxx gx_jconv_settings.cpp -> gx_jconv_settings_1.o}

[brummer]

it's a pity, okay, I think now I get it, cant test it here with gcc 4.4 , but . . .

updated, try again, . . .

[thorgal]

looks weird to me.
I compiled svn@257 with g++ 4.4, no glitch at all. This gcc stuff is highly irritating, from one minor release to the next, they reshuffle basic system headers and break gazillions of project compilation. Not good.

EDIT: fixed a small initialization logic issue when no jconv stuff was found. Try svn@264

[raboof]

This gcc stuff is highly irritating, from one minor release to the next, they reshuffle basic system headers and break gazillions of project compilation. Not good.

To be fair, they don't change the 'official' header you should include to get a function, do they? If it worked before, that was more or less by accident because some header you did include used to depend on some header you used without including it.

Still irritating, though, of course - i've been bitten by it several times also.

[brummer]

they change the internal includes from the headers, so at least you need to include every single header you need to be prevent for thuch a "bug".

But how do that ? If you look around you found several projekts bitten by that every new gcc release.

Anyway, I guess for gcc 4.4 we are now on the run
Let's see what hapen when the next gcc version arive.
For gcc 4.4 they have list here the changes:
http://gcc.gnu.org/gcc-4.4/porting_to.html

for other versions simply change the version number in the adress.

@thorgal yup, resampling works now nice again.

[raboof]

you need to include every single header you need to be prevent for thuch a "bug".

That doesn't sound entirely unreasonable.

But how do that? If you look around you found several projekts bitten by that every new gcc release.

Perhaps there should be a lint-like tool that check whether you're not making assumptions you shouldn't be making.. I don't know of such a tool though.

[thorgal]

here we go:

C++ language issues
Header dependency changes

Some of the standard C++ library include files have been edited to include only the smallest possible number of additional files. As such, C++ programs that used std::printf without including <cstdio>, or used uint32_t without including <stdint.h> will no longer compile.

In detail:

The file <cstdio> is no longer included as part of <string>, <ios>, <iomanip>, <streambuf>, or <locale>.

The file <stdint.h> is no longer included as part of <string> or<ios>.

@brummer: I guess we can do something I wanted to avoid:
gather all system headers in guitarix.h at the top of the file, including the statement "using namespace std;" So we can unclutter the cpp files. I don't like this, I like to see the explicit dependencies in each source unit to be turned into an object. But this gcc crap tends to create too much overhead each time the gcc folks are breaking something (of course, they break it for the better ... )


@raboof: there are tools out there, not sure they're free. I used to use something called flexlint on solaris. There's also a bigger tool called klockwork. Don't know if they are available for free on linux. Maybe some eclipse modules can do the job ? I am not familiar with eclipse but it looks quite exhaustive from what I've seen at work (some colleagues use it for all sorts of things, alsoas a bridge interface to manipulate clearcase vobs elements from some clearcase view storage).

[brummer]

@brummer: I guess we can do something I wanted to avoid:
gather all system headers in guitarix.h at the top of the file, including the statement "using namespace std;" So we can unclutter the cpp files. I don't like this, I like to see the explicit dependencies in each source unit to be turned into an object. But this gcc crap tends to create too much overhead each time the gcc folks are breaking something (of course, they break it for the better ... )

Don't know, I realy like the new style you have bring to the guitarix source, may it's a bit more work when gcc changed some thing, but this way (include depends in every object) it's better to controll what we need for etch single step.

So I mean let us bite in the soure apple, watch out when a new gcc released, and refuel the "new needs".