[Bioc-devel] seq*apply deprecation

Hi,


It may be time to deprecate the seq*apply family of functions, including
seqapply, mseqapply, tseqapply, seqby and seqsplit. These are ancient
functions named in referece to the "Sequence" class, now called "Vector".
It would seem better for the user to write: asList(lapply(...)) instead of
coming up with an alias for that. For readability, we should probably
export asList() as an alias for as(x, "List"), which I just cleaned up.

There is one caveat though: mseqapply has been particularly useful, because
it delegates to mapply_List, which is smart enough to handle any vector
argument in "...". The mapply,List method of course restricts everything to
List. We should probably export mapply_List, probably with a better name.
Suggestions?

Michael

	[[alternative HTML version deleted]]

On Wed, Jan 22, 2014 at 1:29 PM, Herv? Pag?s <hpages at fhcrc.org
<mailto:hpages at fhcrc.org>> wrote:

    Hi Michael,


    On 01/12/2014 05:16 AM, Michael Lawrence wrote:

        Hi,


        It may be time to deprecate the seq*apply family of functions,
        including
        seqapply, mseqapply, tseqapply, seqby and seqsplit. These are
        ancient
        functions named in referece to the "Sequence" class, now called
        "Vector".
        It would seem better for the user to write: asList(lapply(...))
        instead of
        coming up with an alias for that. For readability, we should
        probably
        export asList() as an alias for as(x, "List"), which I just
        cleaned up.


    Sounds good to me.



        There is one caveat though: mseqapply has been particularly
        useful, because
        it delegates to mapply_List, which is smart enough to handle any
        vector
        argument in "...". The mapply,List method of course restricts
        everything to
        List. We should probably export mapply_List, probably with a
        better name.
        Suggestions?


    Why do we need all this?


Neat. Presumably due to this patch:

commit 42139699d91a34c401eedd36db62716a607d0b06
Author: maechler <maechler at 00db46b3-68df-0310-9c12-caf00c1e9a41>
Date:   Wed Nov 28 10:31:42 2012 +0000

     mapply(F, <S4>) now works thanks to a version of Herve's patch

We should do away with mapply_List now that this is part of released R.

       x <- RleList(11:14, 3:1)

    Then:

       > IRanges:::mapply_List(__function(a, b) {a+b}, x, 10)

       [[1]]
       numeric-Rle of length 4 with 4 runs
         Lengths:  1  1  1  1
         Values : 21 22 23 24

       [[2]]
       numeric-Rle of length 3 with 3 runs
         Lengths:  1  1  1
         Values : 13 12 11

    But also:

       > base::mapply(function(a, b) {a+b}, x, 10)

       [[1]]
       numeric-Rle of length 4 with 4 runs
         Lengths:  1  1  1  1
         Values : 21 22 23 24

       [[2]]
       numeric-Rle of length 3 with 3 runs
         Lengths:  1  1  1
         Values : 13 12 11

    What am I missing?

    Thanks,
    H.


        Michael

                 [[alternative HTML version deleted]]

Sounds good. But who is "she"?

On Wed, Jan 22, 2014 at 3:48 PM, Herv? Pag?s <hpages at fhcrc.org
<mailto:hpages at fhcrc.org>> wrote:

    On 01/22/2014 03:02 PM, Michael Lawrence wrote:




        On Wed, Jan 22, 2014 at 1:29 PM, Herv? Pag?s <hpages at fhcrc.org
        <mailto:hpages at fhcrc.org>
        <mailto:hpages at fhcrc.org <mailto:hpages at fhcrc.org>>> wrote:

             Hi Michael,


             On 01/12/2014 05:16 AM, Michael Lawrence wrote:

                 Hi,


                 It may be time to deprecate the seq*apply family of
        functions,
                 including
                 seqapply, mseqapply, tseqapply, seqby and seqsplit.
        These are
                 ancient
                 functions named in referece to the "Sequence" class,
        now called
                 "Vector".
                 It would seem better for the user to write:
        asList(lapply(...))
                 instead of
                 coming up with an alias for that. For readability, we
        should
                 probably
                 export asList() as an alias for as(x, "List"), which I just
                 cleaned up.


             Sounds good to me.



                 There is one caveat though: mseqapply has been particularly
                 useful, because
                 it delegates to mapply_List, which is smart enough to
        handle any
                 vector
                 argument in "...". The mapply,List method of course
        restricts
                 everything to
                 List. We should probably export mapply_List, probably
        with a
                 better name.
                 Suggestions?


             Why do we need all this?


        Neat. Presumably due to this patch:

        commit 42139699d91a34c401eedd36db6271__6a607d0b06
        Author: maechler <maechler at 00db46b3-68df-0310-__9c12-caf00c1e9a41>
        Date:   Wed Nov 28 10:31:42 2012 +0000

              mapply(F, <S4>) now works thanks to a version of Herve's patch

        We should do away with mapply_List now that this is part of
        released R.


    Furthermore, maybe she should also remove the mapply() S4 generic from
    the BiocGenerics package. It's not needed anymore. base::mapply() should
    work out-of-the-box on any object for which length() and [[ are
    implemented.

    H.



                x <- RleList(11:14, 3:1)

             Then:

                > IRanges:::mapply_List(____function(a, b) {a+b}, x, 10)

                [[1]]
                numeric-Rle of length 4 with 4 runs
                  Lengths:  1  1  1  1
                  Values : 21 22 23 24

                [[2]]
                numeric-Rle of length 3 with 3 runs
                  Lengths:  1  1  1
                  Values : 13 12 11

             But also:

                > base::mapply(function(a, b) {a+b}, x, 10)

                [[1]]
                numeric-Rle of length 4 with 4 runs
                  Lengths:  1  1  1  1
                  Values : 21 22 23 24

                [[2]]
                numeric-Rle of length 3 with 3 runs
                  Lengths:  1  1  1
                  Values : 13 12 11

             What am I missing?

             Thanks,
             H.


                 Michael

                          [[alternative HTML version deleted]]