Michael G Schwern <[EMAIL PROTECTED]> writes:
>On Thu, Jan 17, 2002 at 10:32:22AM +0000, Nick Ing-Simmons wrote:
>> The main problem seems to be:
>> 
>> /usr/lib/libpt-1.2.0.so.1: undefined symbol: __ti8iostream
>> 
>> Now -lpt is one I have never seen but MakeMaker has been probing for it for
>> years. Seems it needs a C++ library as well.
>
>After yanking the -lpt out as per your instructions, everything works
>fine (still have lots of warnings) 

Are all/most of the warnings pointer/integer size mis-matches in a 64-bit perl?

>for both stock 5.6.1 and 5.6.1 w/MM
>5.48_02.  Yay!
>
>libpt is really 1.2.5 

What is libpt - someone told me to put it there (for AIX???) years ago,
I have never had a machine which found it.

>and libtk is 8.3.

Irrelevant never used - all the tk code is in Tk.so etc.

>
>
>> >so now I
>> >don't know if its MakeMaker or Debian or Linux/PowerPC or Tk or Solar
>> >Flares that's causing the problem.  They all fail in different ways
>> >(see attached files).
>> 
>> Released Tk is known not to build on bleadperl (can't remember why
>> but I have a version which does).
>
>bleadperl@14190 almost works.  One failure:
>
>t/fileevent......Deep recursion on subroutine "Tk::Event::IO::READ" at 
>blib/lib/Tk/Event/IO.pm line 55, <DATA> line 77.
>t/fileevent......dubious                                                     
>        Test returned status 0 (wstat 11, 0xb)

I have a patch for that. bleadperl took out the support (i.e. 
recursion suppression) for self-tied handles which Tk was using.
I now dynamically create and destroy local globs to use instead 
which means error messages are not as nice and it is slower.
If I feel the urge I may implement my own recursion suppression 
by removing the magic before making the core-call then putting it back.

>DIED. FAILED test 1
>        Failed 1/1 tests, 0.00% okay
>
># bleadperl -Mblib -w t/fileevent.t 
>1..1
>Deep recursion on subroutine "Tk::Event::IO::READ" at 
>/home/schwern/.cpan/build/Tk800.023/blib/lib/Tk/Event/IO.pm line 55, <DATA> line 77.
>Segmentation fault
>
>the problem is for some reason an IO::Handle object is calling
>Tk::Event::IO::READ when sysread() is called on it.  Smells like a
>perlbug.

Smells like an "I told you so" ;-)

-- 
Nick Ing-Simmons
http://www.ni-s.u-net.com/

Reply via email to