Re: [Boost-build] building 1_34_1 on AIX with xlC

From: Marty Fried (public.forums_at_[hidden])
Date: 2007-10-11 14:34:24

"Vladimir Prus" <ghost_at_[hidden]> wrote in message 
news:E1Ig0vo-0002qx-3D_at_zigzag.lvk.cs.msu.su...
> Built bjam by:
>
>        ./build.sh <other-options> debug
>
> then, run the above commnad under debugger. When using gdb, that's:
>
>        gdb --args ./tools/jam/src/bin.aixppc/bjam --toolset=vacpp install
>        (gdb) run
>
> but I have no idea what's debugger is on AIX. When it segfaults,
> get and post a backtrace, or maybe even poke at the code to see what's 
> wrong.
> But actually, 'install' is 'stage' + installing headers, so you might want
> to get 'stage' to work first.
>
I think you're right about getting stage to work first.  Do the headers get 
installed before
the compile?

I started over, and compiled regex using --with-regex option successfully. 
Then I tried it
again changing only the stage option to install and it crashed, so I'm 
giving up on that.

Then, I tried --with-filesystem, the library that didn't build previously. 
It got errors and still
does not build.  I get errors about undefined declarations in 
operations.hpp, is_abstract.hpp,
is_convertible.hpp.

For example:
"./boost/filesystem/operations.hpp", line 732.72: 1540-1118 (S) The 
declaration of "end_itr"
uses the undefined class 
"boost::filesystem::basic_directory_iterator<boost::filesystem::
basic_path<std::basic_string<char,std::char_traits<char>,std::allocator<char> 
 >,
boost::filesystem::path_traits> >" when the class must be complete.

I assume there is something broken in the compiler or the environment; I'll 
try to find out
more, but so far, I'm at a loss.

>>
>>>> Next, I used the stage option, and I don't get the core dumps.  I let 
>>>> it
>>>> run
>>>> overnight, and today, xlCentry was stalled after running 681 minutes.
  [snipped]

> Ouch! If compiler hangs compiling Boost, it seems like compiler bug. In 
> fact,
> that particular file is expected to take a long to compile, so maybe you 
> did
> not wait long enough?

Well, I waited 681 minutes.  Usually, it doesn't take that long to compile 
one
module - how long should I wait?  :-)  This is a fairly fast system, and 
usually flies
compared to our other systems.

I'll see if I can do something about the compiler.

Thanks a lot for you help; I know it's not easy to troubleshoot problems on
a system like this.

Marty Fried
Left Coast, USA