[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: ITS: xf86-input-tslib (updated package)



On Sun, 21 Dec 2008 23:32:15 +0900
"Paul Wise" <pabs@debian.org> wrote:

> On Sun, Dec 21, 2008 at 9:37 PM, Cyril Brulebois <kibi@debian.org>
> wrote:
> > Neil Williams <codehelp@debian.org> (19/12/2008):
> >> - Depends: ${shlibs:Depends}, ${misc:Depends}, ${xserver:Depends}
> >> + Depends: libts-0.0-0 (>= 1.0-5), ${shlibs:Depends},
> >> ${misc:Depends}, ${xserver:Depends}
> >
> > That's ugly. Use shlibs.local (Policy §8.6.5) instead.
> 
> If that is nessecary, you have EPIC FAIL. Go back and start again.

I didn't mention it at the time, but tslib uses symbols so there was
little chance that an shlibs bump would be necessary. There were two
options:
1. The build-dep had been bumped for artificial reasons (which turned
out to be true)
2. testing with 1.0-4 had identified bugs (unreported to the BTS) that
disappeared with 1.0-5 due to the patches implemented in 1.0-5. These
patches did not add new symbols or modify any symbols, the patches
affect internal code within existing symbols. This can change
behaviour, (and therefore bugs), but does not mean an shlibs bump. (In
the event, this was not the reason for the build-dep bump.)

I was merely clarifying that if [2] had been the reasoning that merely
altering the build-dep would not have had the desired effect of
preventing these "bugs" from re-appearing in the case of any failed
migration of tslib alongside the xorg driver.

-- 


Neil Williams
=============
http://www.data-freedom.org/
http://www.linux.codehelp.co.uk/
http://e-mail.is-not-s.ms/

Attachment: pgpRWr6rClcfZ.pgp
Description: PGP signature


Reply to: