Segmentation fault in setting signal handlers; version 3.4.2

[John W. Eaton]

On  3-Oct-2011, Yury T. wrote:

| Hello,
| 
| I'm getting a 'Segmentation fault' when trying to run the 3.4.2 version
| compiled from sources.
| 
| The platform is "ancient" Fedora with gcc (GCC) 4.1.2 20070925 (Red Hat
| 4.1.2-33), powering the small AMD64-based cluster.
| 
| Trying to locate the offending operation, I've reached the procedure
| octave_catch_interrupts in src/sighandlers.cc, in which this fails:
| 
|    retval.int_handler = octave_set_signal_handler (SIGINT, sigint_handler);
| 
| in which procedure, this fails:
| 
|   gnulib::sigemptyset (&act.sa_mask);
| 
| >From the archives, I understand this is something of a common thing
| happening. However, I'm not capable of proceeding any further on my own.
| Also, I didn't quite understand whether some definite solution was found for
| this since the last time this was reported
| 
(http://octave.1599824.n4.nabble.com/octave-3-4-0-on-Scientific-Linux-5-td3324135.html#a3324618
| and https://savannah.gnu.org/bugs/?30685).
| 
| Like the guy commenting on savannah, I tried removing (commenting out) those
| three lines, and now octave works (?), but am I not getting myself an
| installation unobviously crippled, this way?
| 
| Any advise, please? Changing the compiler/platform is rather out of question
| at the moment.

I don't really know of a solution other than upgrading the compiler,
since as far as I could ever tell, there was nothing wrong with what
Octave was doing, and the problem only appeared with the old GCC
4.1.2.

jwe