Re: display problems with some apps

[Sebastian Reitenbach]

Hi,

On Saturday, April 09, 2011 07:12:46 pm Fred Kiefer wrote:
> On 09.04.2011 14:37, Sebastian Reitenbach wrote:
> > On 04/09/11 14:31, Fred Kiefer wrote:
> >> On 09.04.2011 13:49, Sebastian Reitenbach wrote:
> >>> It seems to be the case whether using llvm/clang or gcc seems to be the
> >>> difference. I compiled everything I need to start AddressManager and
> >>> FTP (gworkspace is still compiling, but I think this will then also
> >>> work) with gcc. I started FTP about 30 times, and always looked good.
> >>> Also started AddressManager a couple of times, and it looked good.
> >> 
> >> This then looks like a bug in clang/libobjc2 to me. I went through the
> >> output file you send me from compiling with clang and there were a few
> >> bugs detected, which I am going to fix later today, but none of them
> >> would justify the results you are getting. Most of them are printf
> >> format errors. It is great that clang pays extra attention here.
> >> 
> >> You could try to use a different combination of runtime and compiler
> >> to narrow down the real culprit. Not sure whether gcc + libobjc2 or
> >> clang + libobjc are supposed to work.
> > 
> > I tried libobjc2 + gcc, I only exchanged llvm/clang with gcc, this is
> > where it works well. In the meantime GWorkspace compiled to the end, and
> > also looks good.
> 
> I just commited fixes for almost all the warnings produced by clang.
> Could you please try again?
> I don't expect this to fix the ugly display :-(

Ok, I recompiled everything with llvm/clang/libobjc2 and your updates, and the 
ugly display persists. Starting gworkspace, I've seen a lot of output like 
this now, which I haven't seen before:

2011-04-11 07:59:02.759 GWorkspace[4590] styleoffsets ... guessing offsets
2011-04-11 07:59:02.761 GWorkspace[4590] styleoffsets ... guessing offsets
2011-04-11 07:59:02.762 GWorkspace[4590] styleoffsets ... guessing offsets

Also, I've seen that the other times, everything was compiled with -O2. So I 
removed that in the gnustep-make config.make to compile without optimizations, 
but still, the displaying problem persists.

Sebastian