[cwn] Attn: Development Editor, Latest OCaml Weekly News

Tue Oct 29 01:03:13 PDT 2019

Hello

Here is the latest OCaml Weekly News, for the week of October 15 
to 29,
2019.

Sorry for the hiatus last week, I was away with no internet 
access.

Table of Contents
─────────────────

MirageOS 3.6.0 released
Is it possible to use Machine Learning and Deep Learning 
frameworks through OCaml?
Category theory for Programmers book - OCaml flavor
Ocaml-lxc (binding to Linux Containers) - progress report, looking 
for feedback and contributors
Ocaml-protoc-plugin 2.0.0
Old CWN

MirageOS 3.6.0 released
═══════════════════════

  Archive: 
  <https://discuss.ocaml.org/t/mirageos-3-6-0-released/4554/1>

Martin Lucina announced
───────────────────────

  We are pleased to announce the release of MirageOS 3.6.0. This 
  release
  updates MirageOS to support [Solo5] 0.6.0 and later.

  New features:

  • Support for the Solo5 `spt' (sandboxed process tender) target 
  via
    `mirage configure -t spt'. The `spt' target runs MirageOS 
    unikernels
    in a minimal strict seccomp sandbox on Linux `x86_64', 
    `aarch64' and
    `ppc64le' hosts.
  • Support for the Solo5 _application manifest_, enabling support 
  for
    multiple network and block storage devices on the `hvt', `spt' 
    and
    `muen' targets. The `genode' and `virtio' targets are still 
    limited
    to using a single network or block storage device.
  • Several notable security enhancements to Solo5 targets, such 
  as
    enabling stack smashing protection throughout the toolchain by
    default and improved page protections on some targets.  For 
    details,
    please refer to the Solo5 0.6.0 [release notes].

  Additional user-visible changes:

  • Solo5 0.6.0 has removed the compile-time specialization of the
    `solo5-hvt' tender. As a result, a `solo5-hvt' binary is no 
    longer
    built at `mirage build' time. Use the `solo5-hvt' binary 
    installed
    in your `$PATH' by OPAM to run the unikernel.
  • `mirage build' now produces silent `ocamlbuild' output by
    default. To get the old behaviour, run with `--verbose' or set 
    the
    log level to `info' or `debug'.
  • New functions `Mirage_key.is_solo5' and `Mirage_key.is_xen',
    analogous to `Mirage_key.is_unix'.

  Thanks to @hannes for help with the release engineering for 
  MirageOS
  3.6.0.

[Solo5] <https://github.com/Solo5/solo5>

[release notes] 
<https://github.com/Solo5/solo5/releases/tag/v0.6.0>

Is it possible to use Machine Learning and Deep Learning 
frameworks through OCaml?
══════════════════════════════════════════════════════════════════════════════════

  Archive:
  <https://discuss.ocaml.org/t/is-it-possible-to-use-machine-learning-and-deep-learning-frameworks-through-ocaml/4570/1>

Brando90 asked
──────────────

  I was wondering if this was possible at all?

  The reason I am asking is because there has been some attempts 
  at
  doing this but from Python to OCaml like languages. Outlined 
  here:
  <https://coq.discourse.group/t/machine-learning-and-hammers-for-coq/303>

Anton Kochkov replied
─────────────────────

  Yes it is totally possible, see these resources:
  • [OCaml Owl] scientific computing library, in particular 
  [Neural
    Networks] and [Algorithmic Differentiation] chapters of their 
    manual
  • [Deep Learning with OCaml] blog post from Jane Street
  • [OCaml in Data Science] discussion at this forum
  • 
  <https://discuss.ocaml.org/t/will-ocaml-be-a-good-choice-for-writing-high-performance-parallelizable-machine-learning-libraries/558>
    discussion at this forum

  If you are interested in improving the OCaml capabilites in this 
  area,
  I think joining Owl development is the way to go, you can see 
  their
  repositories here: <https://github.com/owlbarn/>

  They have some ideas on possible improvements of the library at
  [Proposed Projects] page.

[OCaml Owl] <https://ocaml.xyz/>

[Neural Networks] <https://ocaml.xyz/chapter/neural.html>

[Algorithmic Differentiation] 
<https://ocaml.xyz/chapter/algodiff.html>

[Deep Learning with OCaml]
<https://blog.janestreet.com/deep-learning-experiments-in-ocaml/>

[OCaml in Data Science]
<https://discuss.ocaml.org/t/ocaml-for-data-science/1878>

[Proposed Projects] <https://ocaml.xyz/project/proposal.html>

Yaron Minsky added
──────────────────

  There's also a couple of followup posts to that first one:

  • One on [reinforcement learning], using pytorch via OCaml 
  (creating
    bots for playing Atari games!)
  • One on [transfer learning] using pytorch as well.

[reinforcement learning]
<https://blog.janestreet.com/playing-atari-games-with-ocaml-and-deep-rl/>

[transfer learning] 
<https://blog.janestreet.com/of-pythons-and-camels/>

Category theory for Programmers book - OCaml flavor
═══════════════════════════════════════════════════

  Archive:
  <https://discuss.ocaml.org/t/category-theory-for-programmers-book-ocaml-flavor/3905/3>

Arul announced
──────────────

  Hello, I've been working on translating this book in OCaml for 
  few
  months now. @mseri has been kind enough to review all of my 
  OCaml
  snippets and provide feedback. As we approach the final stages 
  of this
  book(I'm currently translating snippets in Chapter21) and the 
  book
  shows some advanced code, we were wondering if it'd be better to 
  get
  more eyes on the OCaml snippets. Would anyone else be interested 
  in
  reviewing the OCaml snippets of this book, atleast from Chapter 
  20 and
  out?

  If so, you could join the conversation here -
  <https://github.com/ArulselvanMadhavan/ocaml-ctfp/pull/22> and I 
  will
  add you as reviewer to all future chapter releases.

  Finally, Thanks @XVilka for posting about the book on here, 
  before I
  even knew that it existed, and getting more traction for the 
  book.

Ocaml-lxc (binding to Linux Containers) - progress report, looking 
for feedback and contributors
════════════════════════════════════════════════════════════════════════════════════════════════

  Archive:
  <https://discuss.ocaml.org/t/ocaml-lxc-binding-to-linux-containers-progress-report-looking-for-feedback-and-contributors/4581/1>

Darren announced
────────────────

  I'm writing to share the progress on [ocaml-lxc], check if 
  anyone's
  interested, and also looking for feedback or potential 
  contributions

[ocaml-lxc] <https://github.com/darrenldl/ocaml-lxc>

Description
╌╌╌╌╌╌╌╌╌╌╌

  ocaml-lxc is an OCaml binding to Linux Containers C API

  The high level exposed design aims to largely match the one of
  [go-lxc], but with adjustments to make it more ergonomic and 
  closer to
  idiomatic OCaml

[go-lxc] <https://github.com/lxc/go-lxc>

Progress
────────

  • Low level API
    • Low level internal API (`lxc_c.ml', `lxc_c.mli') which 
    matches
      closely with C API has been completed, and offers (almost)
      complete coverage of the C API
  • High level API
    • Core parts of the high level exposed API (`lxc.ml`, 
    `lxc.mli`)
      have been completed, you can see [here] for feature set 
      comparison
      to go-lxc
  • Documentation
    • You can see [here] for the documentation of the high level 
    API. I
      have not actually written the documention yet, so this is 
      only
      useful for examining the library structure atm.

[here]
<https://github.com/darrenldl/ocaml-lxc/blob/master/doc/GO_LXC_COMPARISON.md>

[here] <https://darrenldl.github.io/ocaml-lxc/>

Architecture
╌╌╌╌╌╌╌╌╌╌╌╌

  This section is more useful if you're planning to contribute. 
  Feel
  free to skip.

  • Code generation
    • Cinaps is used to generate glue code for both C and OCaml 
    side to
      access `lxc_container' C struct function pointer fields. See
      `code_gen/gen.cinaps' for all the core code generation code.
  • Stubs generation
    • Stubs generation via Ctypes is used to offer high confidence 
    that
      things are done correctly at compile time. See
      `code_gen/ffi_bindings.ml', `code_gen/ffi_types.ml',
      `code_gen/types.ml', and `code_gen/stubs_gen.ml'.
  • C glue code
    • Additional glue code is added to handle things like 
    conditional
      compilation based on LXC version, working around anonymous 
      struct
      (not supported by Ctypes), etc. See `code_gen/lxc_glue.c' 
      and
      `code_gen/lxc_glue.h'.
  • Low level API
    • Low level API ties all the loose ends together, provides a 
    fairly
      direct mapping to the C API, and is the main abstraction 
      layer
      used by the higher level API. See `src/lxc_c.ml' and
      `src/lxc_c.mli'.
  • High level API
    • High level API introduces more abstraction, masking C 
    objects, and
      has a more hierarchical structure compared to the flat 
      structure
      used by lower level API. See `src/lxc.ml' and `src/lxc.mli'.

What I'm looking for
╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌

  • If you're interested in using the library, please feel free to
    provide feedback on the API design
    • e.g. what can be improved, what you don't like, what's 
    missing
  • If you're interested in contributing, please feel free to 
  reach out
    to me to coordinate etc. Following is the rough todo list I 
    have atm
    • Add test code (copy go-lxc's test suite)
    • Add examples
    • API documentation
    • Document build flow architecture (specifically how dune fits 
    into
      it maybe)
    • Document code architecture

TL;DR
╌╌╌╌╌

  A lot of the core bits of ocaml-lxc are done and took way too 
  much
  time, looking for people to help out with the remaining really 
  tedious
  yet crucial parts. And get sanity checks from people who may 
  become
  users of ocaml-lxc.

Links
─────

  This is just a aggregated list of the links above
  ocaml-lxc repo
        <https://github.com/darrenldl/ocaml-lxc>
  go-lxc repo
        <https://github.com/lxc/go-lxc>
  Comparison to go-lxc
        <https://github.com/darrenldl/ocaml-lxc/blob/master/doc/GO_LXC_COMPARISON.md>
  API Documentation
        <https://darrenldl.github.io/ocaml-lxc/>

Ocaml-protoc-plugin 2.0.0
═════════════════════════

  Archive:
  <https://discuss.ocaml.org/t/ann-ocaml-protoc-plugin-2-0-0/4582/1>

Anders Fugmann announced
────────────────────────

  It is my pleasure to announce release of 2.0.0 of
  [ocaml-protoc-plugin].

[ocaml-protoc-plugin] 
<https://github.com/issuu/ocaml-protoc-plugin/>

Highlights for this release
╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌

  • Moved to a new name: The main module have been renamed to
    `Ocaml_protoc_plugin' (from `Protobuf'), to avoid conflicts 
    with
    projects still using `ocaml-protoc'
  • Messages with only one field are not wrapped in a record by
    default. This can be controlled by the option 
    `singleton_record'.
  • Oneof structures with only one field are not wrapped in a
    Polymorphic variant (controlled by option `singleton_record')
  • Added googles `well known types'

  The release is available through opam: `opam install
  ocaml-protoc-plugin.2.0.0', or from [github]

[github]
<https://github.com/issuu/ocaml-protoc-plugin/releases/tag/2.0.0>

Full changelog
╌╌╌╌╌╌╌╌╌╌╌╌╌╌

  • Add examples
  • (*) Oneofs with only one element should not be a variant type
  • Add test when including proto files which defines the same 
  package
  • Add google well know types (library
    `ocaml-protoc-plugin.google_types').
  • (*) Move module to ocaml-protoc-plugin
  • Optimize deserialization of large nested structures
  • Provide pretty_printers aka deriving_show for `Result.error' 
  and
    `Field.t'
  • Fix stack overflow when deserializing big nested structures
  • (*) Add option to not wrap single field type in records
  • Refactor type emitter to closely follow spec (refactor)

  (*) indicates breaking change

Old CWN
═══════

  If you happen to miss a CWN, you can [send me a message] and 
  I'll mail
  it to you, or go take a look at [the archive] or the [RSS feed 
  of the
  archives].

  If you also wish to receive it every week by mail, you may 
  subscribe
  [online].

  [Alan Schmitt]

[send me a message] <mailto:alan.schmitt at polytechnique.org>

[the archive] <http://alan.petitepomme.net/cwn/>

[RSS feed of the archives] 
<http://alan.petitepomme.net/cwn/cwn.rss>

[online] <http://lists.idyll.org/listinfo/caml-news-weekly/>

[Alan Schmitt] <http://alan.petitepomme.net/>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.idyll.org/pipermail/caml-news-weekly/attachments/20191029/28590705/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 487 bytes
Desc: not available
URL: <http://lists.idyll.org/pipermail/caml-news-weekly/attachments/20191029/28590705/attachment-0001.pgp>