DOC: np.random.random documentation refers only to np.random.random_sample #7512

Activity

1. mattip

   added

   04 - Documentation

   component: numpy.random

   on Jul 19, 2018
2. mattip

   changed the title [-]doc error at https://docs.scipy.org/doc/numpy/reference/generated/numpy.random.random.html#numpy.random.random[/-] [+]DOC: np.random.random documentation refers only to np.random.random_sample[/+] on Aug 8, 2018
3. streeck commented on Oct 31, 2018

   

   streeck

   on Oct 31, 2018

   More actions

   Hi @keithbriggs, I've submitted a PR for this, could you have a look and see if it's ok?
4. keithbriggs commented on Nov 3, 2018

   

   keithbriggs

   on Nov 3, 2018 via email

   Author

   More actions

   Thanks for this; looks ok to me.

   …

   ________________________________________ From: Charles David [[email protected]] Sent: 31 October 2018 23:14 To: numpy/numpy Cc: Briggs,KM,Keith,TUD2 R; Mention Subject: Re: [numpy/numpy] DOC: np.random.random documentation refers only to np.random.random_sample (#7512) Hi @keithbriggs<https://github.com/keithbriggs>, I've submitted a PR for this, could you have a look and see if it's ok? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#7512 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AE6IEcFH0y-bfF7aUV25I7tuw7tKqlMrks5uqi7egaJpZM4IANv2>.
5. ron819 commented on Nov 15, 2018

   

   ron819

   on Nov 15, 2018

   More actions

   If someone going to pick this one up.
   Some comments are written in the previous PR:
   #12303
6. bashtage commented on Apr 11, 2019

   

   bashtage

   on Apr 11, 2019

   Contributor

   More actions

   @mattip These legacy aliases should be dropped in randomgen. Both random and sample, and only random_sample should remain.
7. mattip commented on Apr 12, 2019

   

   mattip

   on Apr 12, 2019

   Member

   More actions

   That sounds like a deprecation and so must undergo a deprecation cycle
8. rgommers commented on Apr 12, 2019

   

   rgommers

   on Apr 12, 2019

   Member

   More actions

   That's not how it sounds to me. "randomgen" is brand new, so nothing needs deprecating there. And the existing API should not be touched as part of introducing randomgen.
9. mattip commented on Apr 12, 2019

   

   mattip

   on Apr 12, 2019

   Member

   More actions

   "legacy aliases"
10. rgommers commented on Apr 12, 2019

    

    rgommers

    on Apr 12, 2019

    Member

    More actions

    "in randomgen" (= new)

    It may be ambiguous. If the suggestion was really to deprecate np.random.random, then no - that's definitely a no-go. It's used way too widely, and there's very little to gain from such a deprecation.
11. bashtage commented on Apr 12, 2019

    

    bashtage

    on Apr 12, 2019 via email

    Contributor

    More actions

    There are no plans to depreciate or remove random/RandomState. In fact, I checked and these are already gone from RandomGenerator.

    …

    On Fri, Apr 12, 2019, 09:17 Ralf Gommers ***@***.***> wrote: "in randomgen" (= new) It may be ambiguous. If the suggestion was really to deprecate np.random.random, then no - that's definitely a no-go. It's used way too widely, and there's very little to gain from such a deprecation. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#7512 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AFU5Reb_F4CDjiVMWefju3q2FpAr7Wr_ks5vgEERgaJpZM4IANv2> .
12. rgommers commented on Apr 12, 2019

    

    rgommers

    on Apr 12, 2019

    Member

    More actions

    This bug report and discussion is about the functions in the numpy.random namespace though, not the legacy or new RandomGenerator.

    To address this problem, we should just fix the np.random.random docstring. Just:

    def random(...):
        """"
        This is an alias of `random_sample`. See the `random_sample` docstring for full documentation.
        """
        random_sample(...)
    

    We do this for other aliases as well.
13. bashtage commented on Apr 12, 2019

    

    bashtage

    on Apr 12, 2019

    Contributor

    More actions

    @rgommers This was mostly to link to @mattip 's issue and to be sure that these legacy names don't make it into the new API.
14. rgommers commented on Apr 12, 2019

    

    rgommers

    on Apr 12, 2019

    Member

    More actions

    and to be sure that these legacy names don't make it into the new API.

    Indeed, that would be good. New API = new chance to make things clean and consistent.
15. bashtage commented on Apr 12, 2019

    

    bashtage

    on Apr 12, 2019

    Contributor

    More actions

    This is (or will be soon) fixed in #13163 using simple pass-throughs that explain these are aliases.
16. mattip

    closed this as completedin #13688on Jun 2, 2019