chicken-users
[Top][All Lists]
Advanced

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

Re: [Chicken-users] [ANN] CHICKEN 4.13.0 release candidate available


From: Peter Bex
Subject: Re: [Chicken-users] [ANN] CHICKEN 4.13.0 release candidate available
Date: Wed, 6 Dec 2017 09:09:11 +0100
User-agent: NeoMutt/20170113 (1.7.2)

On Wed, Dec 06, 2017 at 12:46:11AM +0100, Thomas Chust wrote:
> Hello,
> 
> some results from the systems I use:

Hi Thomas,

Thanks for testing!

> Operating system: Linux Mint 17.3 / Ubuntu 14.04 / Linux 3.19.0-80-generic
> Hardware platform: i686
> C Compiler: i686-w64-mingw32-gcc (GCC) 4.8.2
> Installation works?: no (compiler error: alloca.h: No such file or directory)
> Tests work?: no
> Installation of eggs works?: no

Is this a cross compilation toolchain?  Has this build worked before?
If it's a regression, we should probably (try to) fix it, but if it's
really a new platform we should try to add it to the chicken-5 branch.

> Operating system: ReactOS 0.4.6
> Hardware platform: i686
> C Compiler: Digital Mars Compiler Version 8.42n
> Installation works?: no (compiler error: unistd.h is not for Windows use)
> Tests work?: no
> Installation of eggs works?: no

Same question as above; did the Digital Mars compiler ever work?
I'm assuming it behaves differently enough from other C compilers to
require extra #ifdef checks in the chicken.h file.

> Operating system: ReactOS 0.4.6
> Hardware platform: i686
> C Compiler: gcc (tdm-1) 4.9.2
> Installation works?: yes
> Tests work?: no (compiler error: (get) bad argument type - not a symbol: 0.0)
> Installation of eggs works?: no (compiler error: (get) bad argument type - 
> not a symbol: 0.0)
> 
> Apparently the build result using the MinGW tool chain on ReactOS has some 
> problem in the reader:
> 
> > csi -p +
> 0.0

This is likely an issue with us relying on libc's strtod(), if ReactOS's
libc reads "+" as 0.0 we see it as a number (we try to read any
s-expression as a number first and if that fails it's treated as a
symbol).  It has come up several times before, and this has been properly
fixed in CHICKEN 5, which does not rely on strtod anymore (except for
decoding embedded flonum literals in binaries).

We could add yet another special case to fix this in CHICKEN 4, but I'm
not sure it's worth the extra complexity (and potential new bugs).  This
probably also means we've never really supported ReactOS in the past.

Cheers,
Peter

Attachment: signature.asc
Description: PGP signature


reply via email to

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