Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 24 Jul 2012 19:01:05 -0400
From:      David Schultz <das@FreeBSD.ORG>
To:        Steve Kargl <sgk@troutmask.apl.washington.edu>
Cc:        svn-src-head@FreeBSD.ORG, Alexey Dokuchaev <danfe@FreeBSD.ORG>, src-committers@FreeBSD.ORG, svn-src-all@FreeBSD.ORG, Steve Kargl <kargl@FreeBSD.ORG>
Subject:   Re: svn commit: r238722 - in head/lib/msun: . ld128 ld80 man src
Message-ID:  <20120724230105.GA68050@zim.MIT.EDU>
In-Reply-To: <20120724205351.GA65323@troutmask.apl.washington.edu>
References:  <201207231913.q6NJDucB040333@svn.freebsd.org> <20120724084335.GB28038@FreeBSD.org> <20120724174206.GA63841@troutmask.apl.washington.edu> <20120724175712.GA66863@zim.MIT.EDU> <20120724205351.GA65323@troutmask.apl.washington.edu>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Jul 24, 2012, Steve Kargl wrote:
> On Tue, Jul 24, 2012 at 01:57:12PM -0400, David Schultz wrote:
> > On Tue, Jul 24, 2012, Steve Kargl wrote:
> > > On Tue, Jul 24, 2012 at 08:43:35AM +0000, Alexey Dokuchaev wrote:
> > > > On Mon, Jul 23, 2012 at 07:13:56PM +0000, Steve Kargl wrote:
> > > > > Author: kargl
> > > > > Date: Mon Jul 23 19:13:55 2012
> > > > > New Revision: 238722
> > > > > URL: http://svn.freebsd.org/changeset/base/238722
> > > > > 
> > > > > Log:
> > > > >   Compute the exponential of x for Intel 80-bit format and IEEE 128-bit
> > > > >   format.  These implementations are based on
> > > > >   
> > > > >   PTP Tang, "Table-driven implementation of the exponential function
> > > > >   in IEEE floating-point arithmetic," ACM Trans. Math. Soft., 15,
> > > > >   144-157 (1989).
> > > > 
> > > > I believe some ports could benefit from OSVERSION bump for this one.
> > > > 
> > > 
> > > I've never done a OSVERSION bump, so you'll need to tell me how.
> > > But, more importantly, I can find no information in the Developer's
> > > Handbook and only two rather terse references in the Porter's
> > > Handbook.  So, what is OSVERSION?  Why do you think it needs a
> > > bump?
> > 
> > It is the same as the __FreeBSD_version bump I mentioned in my
> > email last week.  Basically it is a number you increment in
> > sys/sys/param.h whenever there is a significant change that
> > porters and developers of third-party software might want to test
> > against.  In this case, it would help any ports that have
> > workarounds for the lack of expl() to compile both before and
> > after this change.  But it's also important not to bump the
> > version gratuitously if there's no reason to believe the change
> > might introduce incompatibilities.
> > 
> > The purpose of each __FreeBSD_version bump is documented here:
> > 
> >   http://www.freebsd.org/doc/en/books/porters-handbook/freebsd-versions.html
> > 
> > We should probably talk about how to update this file at some
> > point, although for the first one or two times, it's probably fine
> > to get a doc committer to help out with this step.
> 
> The only time that __FreeBSD_version has been  bumped for a libm change
> occurred at
> 
> 802502 	March 6, 2011	  	8.2-STABLE after merging log2 and log2f into libm.
> 900027 	December 5, 2010 	9.0-CURRENT after the addition of log2 to libm.
> 
> Conspicuously, missing are bumps for additions of sqrtl(), cbrtl(), long
> double trig. functions, exp2l(), and few other changes.  But, if people
> want a bump here's a diff

I believe there's at least one other; if a bump occurred due to
another recent change, we usually don't bump it again.  For most
of those functions there weren't any workarounds in ports that had
to be ifdef'd.  With expl(), there probably are, although I
haven't checked.

> Index: sys/sys/param.h
> ===================================================================
> --- sys/sys/param.h     (revision 238752)
> +++ sys/sys/param.h     (working copy)
> @@ -58,7 +58,7 @@
>   *             in the range 5 to 9.
>   */
>  #undef __FreeBSD_version
> -#define __FreeBSD_version 1000015      /* Master, propagated to newvers */
> +#define __FreeBSD_version 1000016      /* Master, propagated to newvers */
>  
>  /*
>   * __FreeBSD_kernel__ indicates that this system uses the kernel of FreeBSD,

This looks good.  Note that if you MFC the change, there will need
to be a different bump in -STABLE.

> I can't change the porters-handbook, as I grab sources with csup and
> the following supfile:
> 
> *default host=cvsup3.FreeBSD.org
> *default base=/usr/home/kargl/freebsd/ncvs
> *default prefix=/usr/home/kargl/freebsd/ncvs
> *default release=cvs
> *default delete use-rel-suffix
> *default compress
> doc-all
> www
> 
> A check out of doc/ yields a file named 
> doc/en_US.ISO8859-1/books/porters-handbook/book.sgml 
> 
> The last entry in this file reads
> 
> 	    <row>
> 	      <entry>1000012</entry>
> 	      <entry>May 2, 2012</entry>
> 	      <entry>10-CURRENT jemalloc import
> 		(rev <svnref>234924</svnref>).</entry>
> 	    </row>

That's the right file, and you can commit changes with
'cvs -dkargl@ncvs.FreeBSD.ORG:/home/dcvs' even if you checked
the file out from a local repo.  As a src committer, you want
to be careful if you touch other repositories, i.e., ask a doc
committer to review it if you're unsure.

> If one checks the URL you posted, one see that
> 1000013 		May 22, 2012 	10-CURRENT after byacc import (rev 235788).
> 1000014 		June 27, 2012 	10-CURRENT after BSD sort becoming the default sort (rev 237629).
> 1000015		 	July 12, 2012 	10-CURRENT after import of OpenSSL 1.0.1c (rev 238405).
> (not changed) 	July 13, 2012 	10-CURRENT after the fix for LLVM/Clang 3.1 regression (rev 238429).
> 
> So, I have no idea were the most current source lives.

Yeah, often people neglect to document version bumps until someone
cleans up after them.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20120724230105.GA68050>