chicken-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Chicken-users] Chicken on Solaris 9


From: Bruce Hoult
Subject: Re: [Chicken-users] Chicken on Solaris 9
Date: 04 Jul 2003 12:17:38 +1200

> > - missing dlopen et al when linking .libs/libchicken.so.0.0.0
> > 
> >   dlopen is in libdl, configure is detecting this and adding -ldl
> >   to SHLIBS, but the linker command line doesn't contain -ldl.
> > 
> >   For now I've modified configure.in to add -ldl to LIBS, but this
> >   doesn't seem to be the correct way to do it.
> 
> That's correct.

OK. I've now also checked out the latest cvs and see that this (and the
next one) were already fixed.  But I have to build 1.10 or later before
I can build cvs, right?

I see that the cvs version removes the -lm as well as adding the -ldl.

 
> > - missing nanosleep when linking .libs/libchicken.so.0.0.0
> > 
> >   I've added to configure.in:
> >   AC_CHECK_LIB(rt, nanosleep, [LIBS="$LIBS -lrt"]) 
> 
> Oops, thanks.

Turns out this was already fixed in cvs too, but using:

AC_SEARCH_LIBS(nanosleep, rt)

I don't know enough about autoconf to know whether one is better
than the other.


> > - lots of warnings for implicit declration of alloca
> >   - done nothing as yet 
> 
> Hm. You get this with gcc, right?

right. 3.2.3, the official compiler for this project (not 3.3)


I'm not sure that alloca is a good idea on most RISC architectures. I
think it often ends up actually allocating the memory in a linked list
on the heap with some magic that keeps track of what the stack pointer
was when the memory was allocated and frees the memory the next time
alloca() is called with the stack the same size or smaller.  So you
get the convenience of not having to free it yourself, but not the
efficiency of stack allocation.

How this interacts with Baker's scheme may be interesting. (5 seconds
thought says it'll probably be OK)

-- Bruce






reply via email to

[Prev in Thread] Current Thread [Next in Thread]