Ok, then maybe we have a way of reproducing your problem too. :-) --with-relocatable-dumped-modules isn't really supported, but it would be nice if it worked of course.

/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)

Previous text:

2003-02-21 12:18: Subject: Re: Failed tests again (v7.4.15/CVS 2003-02-21)

---

On Fri, Feb 21, 2003 at 11:35:03AM +0100, Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) @ Pike (-) developers forum wrote:

...

Why are you configuring with "--with-relocatable-dumped-modules"? Does it help if you don't?

Wow... Indeed - it does help :) Many thanks :)

Regards, /Al

/ Brevbäraren

BTW, why -O2 in OPTIMIZE is always added at the end? I can't specify -O3 (-O2 takes precedence because it is last). Well, actually I can override it, but the default behavior... Is strange... :)

It shouldn't matter. I made a routine a while back that used the first occurance of -O<n> and filtered away the rest. How do you feed -O3 to configure? CFLAGS?

/ Mirar

Previous text:

2003-02-21 18:13: Subject: Re: Failed tests again (v7.4.15/CVS 2003-02-21)

---

On Fri, Feb 21, 2003 at 06:00:03PM +0100, Mirar @ Pike developers forum wrote:

...

I doubt it's *because* the Calendar module. :)

Well, tests failed on Calendar modules, so... :)

BTW, why -O2 in OPTIMIZE is always added at the end? I can't specify -O3 (-O2 takes precedence because it is last). Well, actually I can override it, but the default behavior... Is strange... :)

Regards, /Al

/ Brevbäraren

Funny, if I add --with-cflags=o3, I get:

OPTIMIZE: -O3 -pipe -fomit-frame-pointer

But wait, you run 7.4. My changes were probably to 7.5 only. So, anyway. I can say "fixed in 7.5". :)

/ Mirar

Previous text:

2003-02-21 18:23: Subject: Re: Failed tests again (v7.4.15/CVS 2003-02-21)

---

On Fri, Feb 21, 2003 at 06:20:01PM +0100, Mirar @ Pike developers forum wrote:

...

occurance of -O<n> and filtered away the rest. How do you feed -O3 to configure? CFLAGS?

CFLAGS or --with-cflags - same result. -O2 is added at the end.

/Al

/ Brevbäraren

It replaces the absolute path prefix of the installed pike with a magic string in all dumped files, so that the installation can be moved to a different directory without redumping.

/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)

Previous text:

2003-02-21 17:57: Subject: Failed tests again (v7.4.15/CVS 2003-02-21)

---

I doubt it's *because* the Calendar module. :)

What does --with-relocatable-dumped-modules do, anyway?

/ Mirar

Indeed. It's not the first time either, IIRC. :)

/ Mirar

Previous text:

2003-02-21 19:48: Subject: Failed tests again (v7.4.15/CVS 2003-02-21)

---

Well, the Calendar module certainly is good at triggering odd problems in the resolver and module dumper. For instance, right now there's some kind of leak involving objects created from __null_program (used interally in the resolver) which only is trigged by the Calendar module.

I've never gathered the strength to dig into that module to figure out what it's doing. :\

/ Martin Stjernholm, Roxen IS