Bug 17848

Summary:

Patches to remove support for CSRG libm

Product:

Base System

Reporter:

Steven G. Kargl <kargl>

Component:

misc

Assignee:

Bruce Evans <bde>

Status:

Closed FIXED

Severity:

Affects Only Me

Priority:

Normal

Version:

5.0-CURRENT

Hardware:

Any

OS:

Any

Attachments:

Description	Flags
file.diff	none
file.diff	none

Description Steven G. Kargl 2000-04-07 19:20:00 UTC

Compilation of CSRG libm is broken.  From the cvs repository,
it appears as if it has been broken for a long, long, long time. 

The SunPro math library in lib/msun is the default library, and
it appears to be maintained.

Fix: cd /usr/src/lib/libm
rm -rf *
cvs delete -m 'Moving libm to attic' .
cvs commit .

Apply this patch to etc/defaults/make.conf

Apply this patch to lib/Makefile
How-To-Repeat: cd /usr/lib/libm
make clean
make cleandepend
make depend
make

Comment 1 joe 2000-05-02 15:19:03 UTC

I've removed the build support for libm, but have left the source
in the tree for the moment.  Bruce Evans says that there's still
some functionality that needs to be merged into msun from libm
first.  Once this has been done msun could do with renaming back
to libm, or libmsun.

Joe

Comment 2 Steven G. Kargl 2000-05-02 17:11:31 UTC

Joe Karthauser said:
> I've removed the build support for libm, but have left the source
> in the tree for the moment.  Bruce Evans says that there's still
> some functionality that needs to be merged into msun from libm
> first.  Once this has been done msun could do with renaming back
> to libm, or libmsun.

Thanks Joe.

I went hunting for the msun sources at Sun's web site/FTP 
server to see if newer sources were available, but came
up empty.

I suspect that bde is the only person who has a functioning
libm.  As far as the functionality in libm that is missing
in msun, I would be interested in some details because I removed
libm several weeks ago and I haven't noticed any problems
with my computations.  Additionally, a quick glance at 
msun/Makefile suggests that the sources in libm are not
used in building the math library.

-- 
Steve

Comment 3 joe 2000-05-02 17:48:35 UTC

On Tue, May 02, 2000 at 09:11:31AM -0700, Steven G. Kargl wrote:
> I suspect that bde is the only person who has a functioning
> libm.  As far as the functionality in libm that is missing
> in msun, I would be interested in some details because I removed
> libm several weeks ago and I haven't noticed any problems
> with my computations.  Additionally, a quick glance at 
> msun/Makefile suggests that the sources in libm are not
> used in building the math library.

For the record, this is what Bruce and Poul-Henning said:

	On Sun, 30 Apr 2000, Poul-Henning Kamp wrote:
  
	> In message <20000430201742.C48739@pavilion.net>, Joe Karthauser writes:
	> >What's your opinion?  Should libm be delegated to the attic, or
	> >should it be made to work again?
  
	Support for building it should be removed.  The sources may as well be
	left around a little longer, at least in -current, until bits from them
	are merged into msun.  The botched source tree (msun instead of libm
	or at least libmsun) will take longer to recover from.
  
	> I wasn't aware that it was broken.  How long time has it been broken ?
  
	Someone removed __pure from <sys/cdefs.h> on 1998/08/24 without updating
	libm.
  
	> if significant amounts of time I think removal is the smartest way.
  
	It's interesting that msun is still missing the gcc-2 version of __pure
	(__pure2) and that the module that was broken (lgamma) is related to
	why I left libm around (gamma is actually gamma in libm and should become
	tgamma (true gamma) in C99-supporting libraries).
  
	Bruce

I'm not familiar enough with the code/concepts to understand what
needs porting.  Bruce, is it worth expanding on above?

Joe

Comment 4 bill fumerola freebsd_committer

2002-03-22 04:59:17 UTC

State Changed
From-To: open->closed

libm has been put to rest

Comment 5 bill fumerola freebsd_committer

2002-03-22 04:59:17 UTC

Responsible Changed
From-To: freebsd-bugs->bde

bde put libm to rest