AW: irrtoolset 5.0.0 release candidate

Michael.Mueller01 at t-systems.com Michael.Mueller01 at t-systems.com
Wed Feb 3 12:43:09 UTC 2010


Hi Nick,

after installing the packages libtool, autoconf, automake and gcc-3.4.6
(from sunfreeware) in our solaris9 environment, i can configure the irrtoolset
(revision 310).

But i get an compile error in the source code of rpslcheck.cc:

	error: `strcasestr`was not declared in this scope

The function strcasestr is define in the gnulib, but i found no way to install
the gnulib on a solaris9 architecture.

I have added the following workarround in rpslcheck.cc:

	char *strcasestr (char *h, char *n)
	{
		char *hp, *np = n, *match = 0;
		
		if(!*np) {
			return hp;
		}
		
		for (hp = h; *hp; hp++) {
			if (toupper(*hp) == toupper(*np)) {
				if (!match) {
					match = hp;
				}
				if(!*++np) {
 					return match;
				}
			} else {
				if (match) { 
					match = 0;
					np = n;
				}
			}
		}
		
 		return NULL; 
	}

and can build the complete irrtoolset.

Also building on a debian (lenny) is possible.

Michael Mueller.



> -----Ursprüngliche Nachricht-----
> Von: rv at kronos.NIC.DTAG.DE [mailto:rv at kronos.NIC.DTAG.DE] Im 
> Auftrag von Ruediger Volk, Deutsche Telekom T-Com - TE141-P1
> Gesendet: Sonntag, 31. Januar 2010 23:25
> An: Nick Hilliard
> Cc: S.P.Zeidler; irrtoolset at lists.isc.org; Mueller, Michael
> Betreff: Re: irrtoolset 5.0.0 release candidate 
> 
> 
>   > On 29/01/2010 22:31, S.P.Zeidler wrote:
>   > > Building requires bison (plain traditional yacc doesn't 
> understand
>   > > %name-prefix or %output, but wants eg yacc -d -prpsl -o 
> rpsl.y.cc rpsl.y,
>   > > then outputs rpsl.y.h which needs to be renamed to 
> rpsl.y.hh in rpsl;
>   > > same with command.y in rtconfig. I doubt it's really 
> worth it working
>   > > around a dependency on bison though.)
>   >
>   > yah, i had a 10 second look at traditional yacc on 
> solaris, but decided
>   > against making it compatible.  Truth is, bison is pretty 
> ubiquitous these
>   > days; and it's also a lot faster than yacc.
> we are traditionally running Solaris - well, as traditional 
> as being stuck with Solaris 9. We are using flex and bison 
> without problems - as far as I understand.  Unfortunately 
> some other problems did prevent us from successfully building 
> the most recent ISC 4.8.*.
> 
>   > > It needs a define of LEX to "flex -l" on NetBSD 5 and prior.
>   > > I discovered a bug in lex on NetBSD-current (which I'm 
> currently hunting)
>   > > but if one fixes the fallout of that bug, irrtoolset 
> compiles and works
>   > > fine.
>   > >
>   > > Further dependencies:
>   > > requires automake and libtool-base
>   >
>   > It's only the SVN source that needs automake;  the 
> released tarball doesn't
>   > need it, although it does require libtool.
>   >
>   > > doesn't require gmake any longer, which is nice(tm)
>   >
>   > yes, although the Makefiles from automate are truly 
> grotesque.  Can't argue
>   > that they work well, but wow, they're ugly.
>   >
>   > I'm not feeling very guilty about putting in dependencies 
> on tools like
>   > bison, flex and libtool.  Most end-users are going to be 
> installing
>   > irrtoolset from a package / port system and not compiling 
> it themselves.
>   > Faidon has a .deb package; I have an updated freebsd port 
> and also a new
>   > macports Portfile ready.  The netbsd port can be fixed up 
> pretty easily.
>   > That just leaves Solaris, which is very much a minority 
> platform anyway.
>   > Haven't done anything about RH and RH derived systems 
> which need rpms.
>   > Anyone interested?
> With new and very experienced help for our software 
> environment, we are checking whether 5.0.0 rc can be built in 
> our Solaris 9 environment without running into serious 
> problems. We'll certainly report results in the next very few 
> days. If we can build 5.0.0 rc with limited effort we'll also 
> later come back with a plan for merging in the extensions 
> that we have done over the past 3 years.
> 
>   > Nick
> 
> Cheers,
>   Ruediger
> 



More information about the irrtoolset mailing list