The wiki roadmap page (http://trac.xapian.org/wiki/RoadMap) suggests
we aim to start a new stable release series every one to two years.
However, we're not doing a good job of implementing that - in fact we're
just about to hit 5.5 years since 1.2.0 was released.
And that's really much too long - we've built up a pile of nice new
features which it's hard for most people to use.
There are still rather a lot of tickets marked for consideration for
this development series - currently 23 earmarked for 1.3.4, 7 for 1.3.4
and 38 for 1.3.x (though 1.3.x is more "suitable for doing in 1.3.x"
than "we want to do for 1.3.x").
I think we need to acknowledge that these aren't going to all get done,
and we're much better off keeping a sensible interval between new
release series - features can then make the next series after they're
actually implemented. If we'd released 1.4.0 in 2012 (or even 2013), we
could have released or be close to releasing 1.6.0 now, and half the new
stuff currently sitting in 1.3.x could have been in general use for 2-3
years already.
I'm going to make a start soon on going through the tickets and
adjusting milestones. Such decisions aren't necessarily final, but
if there's a ticket you feel is a "must get done for 1.4", then
providing a quality patch is the realistic way to make that happen.
Just saying "+1" isn't going to help.
Cheers,
Olly
where is the source hosted? I can not find where to get it from the website. On Mon, Oct 26, 2015 at 6:05 PM, Olly Betts <olly at survex.com> wrote:> The wiki roadmap page (http://trac.xapian.org/wiki/RoadMap) suggests > we aim to start a new stable release series every one to two years. > However, we're not doing a good job of implementing that - in fact we're > just about to hit 5.5 years since 1.2.0 was released. > > And that's really much too long - we've built up a pile of nice new > features which it's hard for most people to use. > > There are still rather a lot of tickets marked for consideration for > this development series - currently 23 earmarked for 1.3.4, 7 for 1.3.4 > and 38 for 1.3.x (though 1.3.x is more "suitable for doing in 1.3.x" > than "we want to do for 1.3.x"). > > I think we need to acknowledge that these aren't going to all get done, > and we're much better off keeping a sensible interval between new > release series - features can then make the next series after they're > actually implemented. If we'd released 1.4.0 in 2012 (or even 2013), we > could have released or be close to releasing 1.6.0 now, and half the new > stuff currently sitting in 1.3.x could have been in general use for 2-3 > years already. > > I'm going to make a start soon on going through the tickets and > adjusting milestones. Such decisions aren't necessarily final, but > if there's a ticket you feel is a "must get done for 1.4", then > providing a quality patch is the realistic way to make that happen. > Just saying "+1" isn't going to help. > > Cheers, > Olly > >-- Jarrod Roberson 678.551.2852
Jarrod writes:> where is the source hosted? I can not find where to get it from the website.Instructions are under the menu entry "Bleeding edge" on the website: * http://xapian.org/bleeding Best regards, Adam -- "People always assume you're some kind of altruist." Adam Sj?gren asjo at koldfront.dk
On Oct 26, 2015, at 10:05 PM, Olly Betts wrote:>And that's really much too long - we've built up a pile of nice new features >which it's hard for most people to use. > >I think we need to acknowledge that these aren't going to all get done, and >we're much better off keeping a sensible interval between new release series >- features can then make the next series after they're actually implemented. > >I'm going to make a start soon on going through the tickets and adjusting >milestones.Hi Olly, This is good news! One of the things we're eagerly awaiting are official Python 3 bindings. Just today we had an Ubuntu Online Summit session about our goal for Ubuntu 16.04 LTS. We are going to drop Python 2.7 from all the 16.04 images. While Python 2.7 will still be available of course, by default you'll only get Python 3.5[*]. As you probably know, we have apt-xapian-index in both Debian and Ubuntu, and because we install that by default, we'll need to port it to Python 3. That of course will require the Python 3 Xapian bindings. In Ubuntu, we have python3-xapian1.3 as a convenience package for folks who want to start the porting effort of their own applications, and I have started to look at porting apt-xapian-index to this version. But python3-xapian1.3 isn't in Debian so it means we'd have to carry a delta in Ubuntu which is suboptimal. The best outcome would be a new upstream Xapian 1.4 with Python 3 bindings, along with an updated Debian package for python3-xapian, which would then just be imported into Ubuntu. Our drop-dead date for removing Python 2.7 is feature freeze on 18-Feb-2016 but as apt-xapian-index is probably the most complex porting task we have, I want to begin this sooner rather than later. Do you think it's realistic to expect a new Xapian release with Python 3 support in Debian say in the next month or so? Do you have any other thoughts about a Python 3 apt-xapian-index? Would it need to be bilingual (i.e. support Python 2 and 3 at the same time) or is a Python 3 only version acceptable? Cheers, -Barry [*] Currently we also have 3.4, but 3.5 is default and we'll soon be dropping 3.4 too. -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 819 bytes Desc: OpenPGP digital signature URL: <http://lists.xapian.org/pipermail/xapian-discuss/attachments/20151104/876b7b3c/attachment.sig>
On Wed, Nov 04, 2015 at 04:17:03PM -0600, Barry Warsaw wrote:> In Ubuntu, we have python3-xapian1.3 as a convenience package for > folks who want to start the porting effort of their own applicationsHmm, I'm not wild about the package descriptions: http://packages.ubuntu.com/source/wily/xapian1.3-bindings "Xapian search engine interface for Python" and "Xapian search engine interface for Python3" don't make it clear that these are packages of an upstream development snapshot, with an API which might change without notice. I'm not clear why the Python 2 bindings from 1.3 have been packaged as well as the Python 3 bindings - that just seems to add to the potential confusion, as there people could easily just see "1.3 > 1.2" and think this is a newer release, and so a better option to be using. Also not happy to see myself listed as the "Original Maintainer" - these packages are really nothing to do with me. Please can you at least improve those descriptions?> Do you think it's realistic to expect a new Xapian release with Python 3 > support in Debian say in the next month or so?We should have a new development snapshot in that timeframe, but at the current rate of progress I can't realistically see 1.4.0 being out that soon. If anyone is keen to see a release sooner, there are certainly ways to help with that. The most obvious is to look at tickets currently marked with one of the "1.3.*" milestones and see whether they can be bumped to a later version, really need to be addressed before 1.4.0, or if there's a way to achieve a partial win with a small amount of work. Cheers, Olly