mxnet-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Tong He <hetong...@gmail.com>
Subject Re: Release process for R
Date Wed, 04 Jul 2018 17:54:32 GMT
Besides, in the end of the front page message from CRAN at
https://cran.r-project.org/ :

" Note that we generally do not accept submissions of precompiled binaries
due to security reasons. All binary distribution listed above are compiled
by selected maintainers, who are in charge for all binaries of their
platform, respectively. "

Best regards,

Tong He

2018-07-04 9:53 GMT-07:00 Chen HY <chenhy12345@gmail.com>:

> According to
> https://cran.r-project.org/web/packages/policies.html#Binary-packages
> “ Policies for when a (Windows or OS X) binary package will be distributed:
> ”
>
> The issue maker Marco mentioned is a Windows user, which might fits CRAN's
> criteria.
>
> 2018-07-04 17:16 GMT+01:00 Naveen Swamy <mnnaveen@gmail.com>:
>
> > Hi Tong,
> >
> > I hear your comments on how R users already build packages themselves,
> but
> > I do think we should provide a easy to install package that comes
> pre-built
> > with all the dependencies(MKL, CUDA, CUDNN for GPU, etc.,) by publishing
> to
> > CRAN, of course we can start by publishing to a S3 bucket while we
> collect
> > feedback from users. I think providing a pre-built package lowers the
> entry
> > barrier to use MXNet-R since building the MXNet backend is a multi-step
> and
> > non-trivial process, I am not sure if we need to R users to know this in
> > the first place.
> >
> > -Naveen
> >
> > On Wed, Jul 4, 2018 at 1:56 AM, Tong He <hetong007@gmail.com> wrote:
> >
> > > Hi Marco,
> > >
> > > There's no barrier on the technical side, as we can statically link
> > > everything and ask Linux R users to download.
> > >
> > > The main reason that we (so far) do not offer the pre-built R package
> for
> > > Linux users is simple: CRAN doesn't pre-build packages and R users are
> > used
> > > to it.
> > >
> > > Some further reasons:
> > > On Windows and OS X, CRAN has its standard environment (compile with
> > Xcode
> > > and MinGW), but on Linux you just have so many options. Users may feel
> > > confused if the pre-built package is configured differently from their
> > > environment.
> > > I am actually not awared of any R package offering a pre-built binary
> for
> > > Linux, also R's `install.packages` doesn't seem to naturally support
> > > installing binary on Linux (no chance to test yet).
> > >
> > > Another potential approach:
> > > There are some R packages having the form as Linux packages. For
> example
> > > the package `rgl` can be installed on ubuntu with `apt-get install
> > > r-cran-rgl`: https://packages.ubuntu.com/xenial/math/r-cran-rgl .
> > > People prefer to install in this way mainly to get rid of difficulties
> in
> > > the dependencies, or in the environment configuration (e.g. for the
> > package
> > > `rJava` one needs to take care of $PATH and the permission of the
> > > installation).
> > >
> > >
> > > Best regards,
> > >
> > > Tong He
> > >
> > > 2018-07-04 1:20 GMT-07:00 Marco de Abreu <
> > > marco.g.abreu@googlemail.com.invalid>:
> > >
> > > > Hi Anirudh,
> > > >
> > > > thanks for that explanation, that's definitely reasonable. Are we
> > > > publishing it as part of the release process or are these independent
> > > > efforts? I have noticed that we don't offer the pre-compiled version
> > for
> > > > download at
> > > > http://mxnet.incubator.apache.org/install/index.html?
> > > > platform=Linux&language=R&processor=CPU
> > > > and thought that we might want to add a link so our users wouldn't
> have
> > > to
> > > > compile it themselves.
> > > >
> > > > Best regards,
> > > > Marco
> > > >
> > > > On Tue, Jul 3, 2018 at 7:45 PM Anirudh Acharya <
> anirudhkrec@gmail.com>
> > > > wrote:
> > > >
> > > > > Hi Marco,
> > > > >
> > > > > A release process for the R-package would probably involve
> publishing
> > > the
> > > > > package to CRAN. Currently it is not done and the r-package would
> > need
> > > > > considerable changes/improvements before it can get published to
> > CRAN.
> > > > >
> > > > > So at present accessing the package from the S3 bucket is the
> surest
> > > way
> > > > to
> > > > > access the R API.
> > > > >
> > > > >
> > > > > Thanks
> > > > > Anirudh
> > > > >
> > > > >
> > > > > On Tue, Jul 3, 2018 at 2:48 AM Marco de Abreu
> > > > > <marco.g.abreu@googlemail.com.invalid> wrote:
> > > > >
> > > > > > Hello,
> > > > > >
> > > > > > do we have a release process for our R frontend? I noticed the
> > issue
> > > at
> > > > > [1]
> > > > > > and it seems like we're only publishing to an S3 bucket which
is
> > not
> > > > > under
> > > > > > Apache. Is there another channel for our users to retrieve that
> > > package
> > > > > or
> > > > > > is this our only supported official way?
> > > > > >
> > > > > > Best regards,
> > > > > > Marco
> > > > > >
> > > > > > [1] https://github.com/apache/incubator-mxnet/issues/10791
> > > > > >
> > > > >
> > > >
> > >
> >
>
>
>
> --
> Chen Hanyang 陈涵洋
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message