As stated earlier, a big -1 from my side on moving to all Python 3.x for 0.14 already. Not because I personally don't think it provides some benefits, but let me explain:
I don't see enough of a benefit of replacing a few np.dot's with the @ syntax or pure keyword based arguments for the few functions we add/change in one version jump. It doesn't make the code faster, more correct or more feature complete. The API will look and work the same from a users perspective from 2.7-3.5. Required features (which are usually rare) from newer numpy/scipy versions can be handled easily inside the respective functions with an exception. In my experience, many of the scikit-image users I know are actually less knowledgable about how to setup a new Python environment from scratch. And it seems the latest Ubuntu is still on 3.4 and Mac also doesn't ship with 3.5 - which leaves out the @ syntax anyway. Also, many of the cluster users are still stuck on 2.x for quite a while. I think, we will lose a significant amount of users without a real benefit. I don't see a compelling reason to lock out X% of users at this point. I can only judge by the feedback in this thread, but 3 out of 5 people seem to have problems with the proposed way forward. It would be nice if there was some way of obtaining installation statistics from PyPI or Anaconda to get a better picture here. The real benefit comes with 3.5, hence I propose to follow the current model to support 2.7-3.x and reevaluate the situation for 0.15. Eventually, 3.5 will be more widespread and in the meantime we do not really sacrifice anything on our side. Doing a find/replace on np.dot/__future__ will work in a year just as fine as it does now. Best, Johannes > On 10 Aug 2016, at 05:17, Nelle Varoquaux <nelle.varoqu...@gmail.com> wrote: > >> On 9 August 2016 at 19:54, Ralf Gommers <ralf.gomm...@gmail.com> wrote: >> >> >> On Wed, Aug 10, 2016 at 2:46 PM, Nelle Varoquaux <nelle.varoqu...@gmail.com> >> wrote: >>> >>>> On 9 August 2016 at 19:17, Ralf Gommers <ralf.gomm...@gmail.com> wrote: >>>> >>>> >>>> On Wed, Aug 10, 2016 at 1:55 PM, Nelle Varoquaux >>>> <nelle.varoqu...@gmail.com> >>>> wrote: >>>>> >>>>> On 9 August 2016 at 17:28, Juan Nunez-Iglesias <jni.s...@gmail.com> >>>>> wrote: >>>>>> @Emmanuelle I'm probably among the ones pushing hardest for this, and >>>>>> I >>>>>> can >>>>>> tell you, I can't wait for this change in policy, and would be sorely >>>>>> disappointed by having to work in an experimental branch of >>>>>> scikit-image. >>>>>> Both @-matmul and keyword-only arguments are, imho, compelling >>>>>> reasons >>>>>> to >>>>>> switch. (Imagine the amount of fiddling with the API that we could do >>>>>> with >>>>>> keyword-only arguments, without the annoying deprecation cycle!) >>>>>> >>>>>> @Ralf I would argue in favour of 3.5, for the above reasons and >>>>>> because >>>>>> anyone who had the temerity to update to 3.4 is very likely to >>>>>> subsequently >>>>>> move to 3.5. (Self-selected group of early adopters, plus no >>>>>> backwards >>>>>> incompatibility issues between the two versions.) >>>> >>>> >>>> I agree with this argument, I was just asking to clarify. @ is at least >>>> something interesting that's specifically added for scientific users, so >>>> imho 3.5 is the first 3.x release where in some cases the benefits may >>>> start >>>> to be worth the costs. >>>> >>>>> >>>>> That implies that even ubuntu users will have to install python from >>>>> another source than the package manager. Do you really want this? That >>>>> means that only fairly advance python users will be able to use the >>>>> latest scikit-image release. >>>> >>>> >>>> In the grand scheme of things, does Ubuntu matter much for this >>>> decision? >>>> There are way more Windows and OS X users, so if it's OK for them (which >>>> is >>>> not a given) then it should also be OK for those fewer and on average >>>> more >>>> computer-literate Ubuntu users. >>> >>> I don't have a good overview of what OS people are using, but in the >>> different research facilities I've worked or visited, it was always >>> linux based, and users where not really tech-savy. I checked on all of >>> the servers (UW's genome science, the Curie institute, UC Berkeley's >>> stats department, the Mines' machine learning computing facilities.) I >>> have access to, none have python3.5. Some of these servers are >>> "offline", thus conda is useless. All of these research institutes >>> have teams that use in some way image processing. >>> >>> I'll just also mentionned that we had a keynote at scipy this year >>> mentionning she was using python because she just did not have the >>> time anymore to code in C++ and Java. These are the kind of people >>> that may be using scikit-image, that may be interested in the latest >>> version of the package, and yet just not have the time to compile >>> python3.5 from scratch on an outdated server. >> >> >> Fair enough, but there are likely much more outdated servers with Python 2.7 >> than with Python 3.4 on them ..... > > Well… I think it is better to keep python2.7 support :) > All the clusters I have access to but 1 has python3. The oldest one is > a red hat based cluster with python2.6 (which was python2.3 not so > long ago). > > N > >> >> Ralf >> >> -- >> You received this message because you are subscribed to the Google Groups >> "scikit-image" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to scikit-image+unsubscr...@googlegroups.com. >> To post to this group, send email to scikit-image@googlegroups.com. >> To view this discussion on the web, visit >> https://groups.google.com/d/msgid/scikit-image/CABL7CQjye5g5jppRk1hej2g2P3VCKa0mrcHe4Z9u37nSrqBpsg%40mail.gmail.com. >> >> For more options, visit https://groups.google.com/d/optout. > > -- > You received this message because you are subscribed to the Google Groups > "scikit-image" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to scikit-image+unsubscr...@googlegroups.com. > To post to this group, send an email to scikit-image@googlegroups.com. > To view this discussion on the web, visit > https://groups.google.com/d/msgid/scikit-image/CAE-UAvT0C5z0jg84oDW0MtTiFkoWJgpdNrMCfGx82ZnwDzLeZg%40mail.gmail.com. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "scikit-image" group. To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image+unsubscr...@googlegroups.com. To post to this group, send an email to scikit-image@googlegroups.com. To view this discussion on the web, visit https://groups.google.com/d/msgid/scikit-image/3FEC278C-8155-4C85-BB44-0B0F7789D071%40demuc.de. For more options, visit https://groups.google.com/d/optout.