[lsb-discuss] qt3: testing our deprecation strategy
Theodore Tso
tytso at mit.edu
Tue Jul 28 07:33:57 PDT 2009
On Mon, Jul 27, 2009 at 11:39:23AM -0600, Wichmann, Mats D wrote:
>
> Our current policy will not let us actually get
> rid of qt3 for quite q while yet. I believe this
> means that there are a raft of forthcoming distributions
> who will not technically be able to comply with LSB 4.0,
> 4.1, etc.
>
> This is not a new topic, but I want to shake the tree
> a little bit on this topic. Qt was kind of a special
> case in LSB, qt3 should never have gone in, but went
> in because qt4 wasn't ready enough yet (by the LSB
> criteria).
Well, perhaps we should think about changing our deprecation strategy.
Originally when we formulated "three major releases", that was during
the LSB 2.x to 3.x time period, if memory serves correctly. At the
time when we made that commitment, I don't think we realized how long
time might stetch between minor releases, let alone major releases.
Indeed, there was an argument that said that LSB 4.0 could have been
named LSB 3.3, since from a specification point of view there were
more changes between 3.1 and 3.2 and 3.2 and 4.0 --- but it had been
over long time since LSB 3.0 had been released (July, 2005), and
people have a tendency to not pay attention at new releases of
specifications or software unless it gets a major version bump.
In any case, it's been 5 years between LSB 3.0 and 4.0, and if that
keeps up, "3 major releases" might be 15 years, which is probably way
too long. Even if we change it to be 3 releases, it may make sense
for us to make exceptions for interfaces/libraries that have never
been used in a certified application. (Which would be a way to
incentivize ISV's to certify their apps. :-)
Maybe, as a strawman proposal, what if we change the deprecation
policy that if the interface/library hasn't been used in a certified
application, an interface can be removed a year after it has been
marked deprecated in an LSB specification? What do folks think?
- Ted
More information about the lsb-discuss
mailing list