F21 System Wide Change: lbzip2 as default bzip2 implementation

[Toshio Kuratomi]

On Thu, Apr 03, 2014 at 04:48:03AM +0200, Zbigniew Jędrzejewski-Szmek wrote:
> On Wed, Apr 02, 2014 at 07:26:59PM -0700, Toshio Kuratomi wrote:
> > On Wed, Apr 02, 2014 at 08:47:11PM -0500, Chris Adams wrote:
> > > 
> > > I think the "right" way to move forward is to make a library that is at
> > > least API-compatible with the current libbz2.so.1, make all the tools
> > > use it, and just replace bzip2 with lbzip2.
> > >
> > Although I'm still on the fence about whether I'd vote for the Change as is,
> > I tend to agree with this sentiment.
> > 
> > Having two sets of code with different characteristics seems like
> > isomething of a disservice to users (I started bzip'ing my logs and backups
> > because the performance was suitable for my task when I tested with
> > /usr/bin/bzip2 but then when I operated on those logs with a custom python
> > script it was 3x slower!)
> > 
> > From past precedent I agree that getting the new package to the point where
> > we think it's a suitable replacement and then just making the switch for the
> > next release makes the most sense.
> I agree that this is desirable. OTOH, lbzip2.rpm is 90k, so I guess we can
> suffer the extra disk usage :)
> 
If it was about disk usage :-)

But it's not.  It's about having two codebases that do the same thing in
different ways.

-Toshio
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <http://lists.fedoraproject.org/pipermail/devel/attachments/20140402/f17f1ae0/attachment-0001.sig>