Page 1 of 3

convincing GiGi to build with other gcc

Posted: Fri Feb 03, 2006 10:11 am
by harm
Hi,

I am going through some pains to get FreeOrion to compile and I actually got scons to pass all the checks but now I have an error like this

Code: Select all

client/human/../../UI/CUIControls.h:64: error: 'CLICKABLE' is not a member of 'GG::Wnd'
Am many more of his friends. Many many more.
But I suspect it has to do with the gcc version I used to compile GiGi(this is an obvious GG error I think). I used 4.0something and it worked but when I was compiling log4cpp 4.0 gave me a hard time and switching back to 3.3 fix the problem.
Is there a way to convince scons to use 3.3 instead of 4.0? I have both installed on my system (debian unstable).
Thanks in advance.

Posted: Fri Feb 03, 2006 12:48 pm
by noelte
Zach is updating GG and now FO too. Maybe you got your code in between cvs updates.

Posted: Fri Feb 03, 2006 1:27 pm
by harm
What do you exactly mean? You think that if I where to update to the newest CVS source this nastly problem goes away? The changes are rather slim that this scenario really happened right?

Posted: Fri Feb 03, 2006 1:39 pm
by harm
I am using the CVS version of GiGi because the official released tar ball 0.6.0 did not(and does not) compile. But I read just about everwhere that I should be using the official release. And again I think the problem is that I tried to compile the official release with gcc version 4.0.x.
So if I can't compile the official release with gcc 4 and I cant use the cvs version of gigi that means I am out of luck? :roll:

Posted: Fri Feb 03, 2006 2:54 pm
by noelte
Guessing based on your error message, i would say GG and FO are out of sync. Getting CVS sources right now might be difficult ...

I don't think the error is related to gcc.

Posted: Fri Feb 03, 2006 3:05 pm
by harm
Aha, I see what you mean. And I indeed noticed that getting the sources now is next to impossible. I keep getting connection refused messages. Why is getting the sources now so difficult?

Posted: Fri Feb 03, 2006 3:10 pm
by noelte
connection refuses are another issue.

Posted: Fri Feb 03, 2006 3:40 pm
by harm
That is just becasue SF is busy right?

Posted: Fri Feb 03, 2006 9:13 pm
by harm
Dammit! Sorry. I did get access to the CVS repositories from both GiGi and FreeOrion. So I downloaded and them, compiled GiGi (with gcc 4.0) and then tried to compile FO. It got further than the first time. So there was a discrepancy between GG and FO. Unfortunately the problem persists. I did notice something while compiling a warning saying I was using libstdc++.so.6 instead of libstdc++.so.6 and that this might lead to problems. After some searching I again noticed that this is a difference between gcc 3.3 and 4.0. Is there any way to get scons to use 3.3 instead of 4.0?

Posted: Fri Feb 03, 2006 9:39 pm
by tzlaine
As of right now, yoiu should be using the latest Subversion files for GG and FO. They are synched up and work for non-Win32 builds. The Win32 fix is in the works now.

Posted: Fri Feb 03, 2006 9:49 pm
by harm
Arrrrr are we using Subversion now? That is great I like it a lot better then CVS but how should I get those sources then? If the CVS and Subversion versions are all the same then I got those build just ~8 hours ago.

Posted: Fri Feb 03, 2006 10:34 pm
by Geoff the Medio
harm wrote:...how should I get those sources then?
You might take a look at:

http://sourceforge.net/svn/?group_id=75752

tzlaine also emailed me this url:

http://svnbook.red-bean.com/nightly/en/ ... svn.forcvs

Posted: Sat Feb 04, 2006 3:22 am
by harm
Aha! I got it, the sources I mean and I even got everything to compile! So after I solved some library files gone missing I typed ./freeorion and...... got a segmentation fault. No other output just that. Are there any other commandline switches I could try?
Thanks anyway for the great help and responsiveness!

Posted: Sat Feb 04, 2006 3:24 am
by harm
Ow the exit code is 139. That might say something.

Posted: Sat Feb 04, 2006 4:14 am
by tzlaine
harm wrote:Ow the exit code is 139. That might say something.
I have no idea what that exit code in particular means. Could you GG and FO rebuild with 'scons debug=1', run it in gdb, and post the stack trace at the segfault?