svn commit: r283969 - head/lib/libutil

[Baptiste Daroussin]

On Thu, Jun 04, 2015 at 12:52:58AM -0700, Xin Li wrote:
> 
> 
> On 6/3/15 22:21, Baptiste Daroussin wrote:
> > On Wed, Jun 03, 2015 at 03:43:46PM -0700, Xin Li wrote:
> >> On 06/03/15 14:58, Baptiste Daroussin wrote:
> >>> On Thu, Jun 04, 2015 at 12:51:46AM +0300, Sergey Kandaurov
> >>> wrote:
> >>>> On 3 June 2015 at 23:48, Baptiste Daroussin
> >>>> <bapt at freebsd.org> wrote:
> >>>>> Author: bapt Date: Wed Jun  3 20:48:28 2015 New Revision: 
> >>>>> 283969 URL:
> >>>>> https://svnweb.freebsd.org/changeset/base/283969
> >>>>> 
> >>>>> Log: Add a pw_mkdb2(3) function which does the same thing
> >>>>> as pw_mkdb(3) except it takes a new argument allowing to
> >>>>> specify the endianness of the database to generate
> >>>>> 
> >>>> 
> >>>> Why not change pw_mkdb()? Is it used outside of the project?
> >>>> 
> >>> Because that would change the ABI of libutil and it is not a 
> >>> private library aka we are supposed to maintain ABI
> >>> compatibility as we do not know if it is used or not externally
> >>> to the project. I care about the ABI because I have made this
> >>> change in order to use it in pw(8) and MFC it to stable/10
> >>> before 10.2.
> >>> 
> >>> libutil is not versionned so this is the only way to not break
> >>> the ABI. Except if someone has a better idea than I do.
> >> 
> >> Looking at r113596, pwd_mkdb(8) was changed to generate both
> >> legacy (version 3, endianness sensitive) and new (version 4,
> >> machine independent) formats.
> >> 
> >> Now, after 12 years, is it still sensible to generate legacy
> >> format db entries?  Maybe we should just disable the generation
> >> by default and eventually remove the ability to generate them?
> >> 
> > 
> > That could be an option, in this case we could add a -l (legacy)
> > option to pwd_mkdb to allow the users to generate the db in legacy
> > format and drop support support for legacy format in all other
> > tools. Meaning I can revert pw_mkdb2(3)
> 
> Well, I think the pw_mkdb2(3) should be reverted, because the base
> system libraries are not using the legacy format when new formatted
> entries are present (which is true for databases generated on FreeBSD
> 5+); and if they do, that shall be seen as a bug and we should fix
> these libraries.
> 
> I have committed r283981-r283983 to implement the default change and
> the -l option.
> 
done, thank you!

Bapt
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/svn-src-head/attachments/20150604/b1956669/attachment.sig>