cvs commit: fptools configure.acfptools/libraries/base/System/Posix Signals.hscfptools/libraries/base/include HsBase.h

[Simon Marlow]

On 07 September 2004 16:54, Ross Paterson wrote:

> On Tue, Sep 07, 2004 at 08:35:45AM -0700, stolz at haskell.org wrote:
>> stolz       2004/09/07 08:35:44 PDT
>> 
>>   Modified files:
>>     .                    configure.ac
>>     libraries/base/System/Posix Signals.hsc
>>     libraries/base/include HsBase.h
>>   Log:
>>   Add cpp-protected signals sigINFO & sigWINCH if available.
>>   (An autoconf-wizard might want to look at the bottom of
>>   configure.ac, the similarities between HAVE_SIGPOLL, HAVE_SIGINFO
>>   & HAVE_SIGWINCH can surely be factored out)
> 
> The issue was already there with SIGPOLL, but:
> 
> - isn't it highly undesirable for a module to have different
>   interfaces in different environments?  (That is, can't these things
>   be exported with value 0?)
> 
> - why is HAVE_SIGPOLL better than #ifdef SIGPOLL?

These shouldn't be in System.Posix.Signals IMHO.  System.Posix is
supposed to implement POSIX, nothing more, nothing less.

I've no objection to putting them *somewhere* - perhaps
System.Posix.Signals.Exts?  With System.Posix.Exts to re-export
non-POSIX things?

I think Ross's objection to HAVE_SIGPOLL is that there's no reason to
autoconf this stuff when we're using hsc2hs.  However, as it is,
System.Posix.Signals shouldn't really be using hsc2hs since it's in
libraries/base (see my recent message on ghc-users).  But it's probably
not a significant problem: the worst that could happen is that ^C
doesn't work properly in stage1 of a newly HC-bootstrapped compiler.  I
need to split System.Posix.Signals into the parts required by GHC, and
the rest which can go in libraries/unix.

Cheers,
	SImon