Re: Qt4 removal plan

[Greg Troxel <gdt%lexort.com@localhost>]

maya%NetBSD.org@localhost writes:

> On Fri, Oct 25, 2019 at 01:29:31PM +0200, Kamil Rytarowski wrote:
>> I propose to obsolete and remove Qt4 from pkgsrc.
>> 
>> Most of the existing users of qt4 packages are KDE4, everything else is
>> dead upstream.
>> 
>> Many of the distributions already did it:
>> 
>> https://wiki.debian.org/Qt4Removal
>> 
>> The last version of Qt4 4.8 LTS (4.8.7) was released in 2011 and is
>> unsupported since 2015.
>> 
>> Qt5 is around since 2012.
>> 
>
>
> I fixed the original reason for suggesting this, namely qt4-libs not
> building on netbsd-current.

I saw that; thanks.

> Can we please do this on a more sensible timeline?

Please see the note I sent on behalf of PMC.

While there have been some comments that talk about timeline, the notion
of a timeline for qt4 removal is outside of pkgsrc guidelines and will
not happen.   Timelines have been used very rarely, after a prolonged
slower process, such as in the destdir conversion process.

Basically:

  It's ok to prune leaf packages that merit removal on their own (as
  unmaintained for a very long time, with no users), following our usual
  process of proposing removal on pkgsrc-users.

  It's not ok to remove things *because* they depend on qt4-libs.  I
  didn't say this clearly, but it's not ok to drop qt4 options without
  the usual unmaintained/no-users/proposal.

  The notion of someone setting timelines for wholesale removal of
  everything that depends on qt4 is not acceptable.  (Maybe I was too
  implicit about this.)

  pkgsrc does not have an intent to prevent people from running
  unmaintained software.  We do not remove things in order to make
  people stop using them.

As I see the status:

  Currently, the presence of qt4 in pkgsrc is not causing any problems.

  Eventually, qt4 will be really untenable.  But as of today, it builds
  and runs pretty much on all pkgsrc platforms.  So the only issue is
  that it is unmaintained upstream.

  This discussion has been useful in that a number of things which could
  be updated to upstream versions that use qt5 have been or are being
  updated.

  A number of things which seem very old and likely to have no users
  have been proposed for removal, and some removed.

  There are some packages using qt4 that have users, and we are
  certainly going to have some number of those.

  KDE4 pretty obviously has users, and any notion of removing it will
  need broad consensus, with no credible objections.  I would be
  interested in hearing from Mark about his view of the  roadmap to
  better KDE5 support and the path to eventually removing KDE4.

Based on the above, I see us doing updates and pruning some cruft, but I
expect a number of qt4-using packages to remain as in use, and I expect
qt4 itself to remain, both as a dependency and because someone wanting
to build software outside of pkgsrc with qt4 seems likely.


Certainly, if anyone sees a specific removal that's objectionable
(including dropping of qt4 options in packages) because there are or are
likely to be users -- speak up and it can be reverted, especially if it
was not properly proposed on pkgsrc-users.