On the other hand signal 9 is SIGKILL, so it might be that some other process kills it due to eg using too much memory.
In the last episode (Jan 03), Henrik Grubbstrm (Lysator) @ Pike (-) developers forum said:
On the other hand signal 9 is SIGKILL, so it might be that some other process kills it due to eg using too much memory.
That's more likely; I remember noticing that peep.c takes a lot of memory to build. gcc 2.9x has major issues with large switch statements and -O. It's got the same problem compiling mysql, on one of its yacc-generated source files. At least the pike build backs down and builds without -O automatically :)
So, what version of GCC does one want to use these days?
/ Marcus Agehall (Trådlös)
Previous text:
2003-01-03 21:01: Subject: Re: Problem compiling Pike
In the last episode (Jan 03), Henrik Grubbstrm (Lysator) @ Pike (-) developers forum said:
On the other hand signal 9 is SIGKILL, so it might be that some other process kills it due to eg using too much memory.
That's more likely; I remember noticing that peep.c takes a lot of memory to build. gcc 2.9x has major issues with large switch statements and -O. It's got the same problem compiling mysql, on one of its yacc-generated source files. At least the pike build backs down and builds without -O automatically :)
-- Dan Nelson dnelson@allantgroup.com
/ Brevbäraren
In the last episode (Jan 03), Mirar @ Pike developers forum said:
3.2.1 might not be recommended since it failed to build a Pike now... (Error 139 in pike when making GL/auto.c, says xenofarm.)
You hit a bad 7.5 checkout; take a look at the number of red lights on the last couple builds. I usually get working pike 7.5's on Solaris 7, Tru64 5.1, and FreeBSD 5 with gcc 3.2.1.
It looks like it is Mirars code that breaks the builds...
/ Martin Nilsson (bygger parser
Previous text:
2003-01-03 23:24: Subject: Re: Problem compiling Pike
In the last episode (Jan 03), Mirar @ Pike developers forum said:
3.2.1 might not be recommended since it failed to build a Pike now... (Error 139 in pike when making GL/auto.c, says xenofarm.)
You hit a bad 7.5 checkout; take a look at the number of red lights on the last couple builds. I usually get working pike 7.5's on Solaris 7, Tru64 5.1, and FreeBSD 5 with gcc 3.2.1.
-- Dan Nelson dnelson@allantgroup.com
/ Brevbäraren
Which is odd, since I haven't changed anything. But I was about to suggest that libmird should be removed from the dist; I'm certain that it's not popular enough to be needed. (The glue can remain, of course.)
/ Mirar
Previous text:
2003-01-03 23:40: Subject: Re: Problem compiling Pike
It looks like it is Mirars code that breaks the builds...
/ Martin Nilsson (bygger parser
It seems that the use of machine code is broken on my system. I'm adding a xenofarm target with --without-machine-code now to see if my theory is correct.
/ Mirar
Previous text:
2003-01-04 07:59: Subject: Re: Problem compiling Pike
Which is odd, since I haven't changed anything. But I was about to suggest that libmird should be removed from the dist; I'm certain that it's not popular enough to be needed. (The glue can remain, of course.)
/ Mirar
pike-devel@lists.lysator.liu.se
-
Dan Nelson -
Henrik Grubbstr�m (Lysator) ＠ Pike (-) developers forum -
Marcus Agehall (Tr�dl�s) ＠ Pike (-) developers forum
-
Martin Nilsson (bygger parser ＠ Pike (-) developers forum
-
Mirar ＠ Pike developers forum
-
Peter Bortas ＠ Pike developers forum