Hello
Here is the latest OCaml Weekly News, for the week of July 04 to 11, 2017.
Archive: https://sympa.inria.fr/sympa/arc/caml-list/2017-07/msg00007.html
Francois BERENGER announced:
I am pleased to announce parany, a kind of minimalistic and more generic version
of parmap.
Yes, a minimalistic version of parmap is possible! :D
Parany can be found here:
https://github.com/UnixJunkie/parany
The super simple interface is:
---
(* the demux function must throw End_of_input once it's done *)
exception End_of_input
val run: nprocs:int ->
demux:(unit -> 'a) ->
work:('a -> 'b) ->
mux:('b -> unit) -> unit
---
This is a beta release, so please don't expect excellent performance and rock
stability. Also, maybe it can crash in some cases (tell me).
The difference with parmap is that parany is supposed to be able
to work with very large files (that you cannot load in memory at once)
or infinite streams of things to process.
Managing such cases is doable with parmap but requires some programming
and is not optimal in terms of parallelization performance (you have to stop all
work when loading in memory a part of your file and
fork all your workers again after that).
Parany relies on the shared data structure Netmcore_queue from
Gerd Stolpmann's excellent ocamlnet library to take care of all the magic.
Contributions are welcome to improve performance and stability, while not
degrading code quality.
Also, since this is a minimalistic library, I am not sure
new features will be accepted. ;)
If there is some interest, I can put it into opam, let me know.
Archive: https://sympa.inria.fr/sympa/arc/caml-list/2017-07/msg00032.html
Francois BERENGER announced:
It is my pleasure to announce the first release of cpm.
cpm allows to compute various classification performance metrics, like:
- area under the ROC curve (AUC)
- BEDROC (Bolzmann Enhanced Discrimination of the ROC curve)
- Power Metric (PM) at given threshold
- Enrichment Factor (EF) at given percentage
Here is an example use:
---
(* first, define your score_label module *)
module SL = struct
type t = string * float * int * bool
let get_score (_, s, _, _) = s
let get_label (_, _, _, l) = l
end
(* second, instantiate the ROC functor for your score_label module *)
module ROC = MakeROC.Make (SL)
(* third, call any classification performance metric you need *)
[...]
let auc = ROC.auc scores in
[...]
---
The score is the output of your predictor.
For a true positive, the label must be 1.
For a true negative, the label must be 0.
Your predictor is supposed to give high scores to true positives
and low scores to true negatives.
The code is here:
https://github.com/UnixJunkie/cpmlib
The library should be available shortly in opam under the name cpm.
For those who like to read, here are some interesting reads on related subjects:
- https://jcheminf.springeropen.com/articles/10.1186/s13321-016-0189-4
- https://academic.oup.com/bioinformatics/article-lookup/doi/10.1093/bioinformatics/btq140
- http://pubs.acs.org/doi/abs/10.1021/ci600426e
Have fun classifying,
F.
Archive: https://sympa.inria.fr/sympa/arc/caml-list/2017-07/msg00010.html
Richard Jones asked:
I have a need to make automated stylistic and syntax checks to
a large amount of OCaml code. The sort of rules would be:
- Check that indentation is used, and used consistently.
- Excessive use of parentheses where not needed.
- Use ( .. ) instead of begin .. end.
- Flag uses of various Obj.* and unsafe_* functions.
It seems to me that some of these could be tested with either camlp4
or ppx. I think if we allowed the checker to go back to the original
code (eg to see if the parser parsed as block as '(' or 'begin'),
it might be able to check all of these things.
Anyway, before I start on it I'm wondering if anyone has ever
looked at doing this kind of thing?
Richard Jones later added:
Now that I actually formulate the right search terms, I see
that ocp-lint exists :-/
Mark Bradley replied:
Another idea that will get you part of the way there is to use refmt to
automatically format the OCaml files to standard layout.
This would cover 3 out of the 4 criteria because:
1. it automatically indents
2. it always reduces parentheses where it can
3. it automatically converts begin and end to parentheses.
The 4th criteria could potentially be caught using grep.
It's worth an experiment to see if you like the style of code generated by
refmt for OCaml. It certainly does a better job at formatting reason code
from my limited experiments (it doesn't seem to like vertical white space
when generating OCaml, whereas it will include it if you covert the file to
reason, testing version 1.13.5 as of writing).
Sébastien Hinderer also replied:
You may also be interested in Xavier Clerc's Mascot tool:
http://mascot.x9c.fr/
Archive: https://sympa.inria.fr/sympa/arc/caml-list/2017-07/msg00013.html
Johannes Kanig continued this thread from last week:
> Here is my list of affected OCaml projects known so far (it seems
> surprisingly less rare than I thought at first):
To add to the list, we at AdaCore have seen random crashes of long-running Why3
processes, that only happen on the Skylake machine of one of our developers.
Archive: https://sympa.inria.fr/sympa/arc/caml-list/2017-07/msg00024.html
Adrien Nader announced:
I am happy to announce the first release of Prose, a collaborative text
editor started after being both frustrated and horrified by Etherpad
Lite.
Etherpad Lite is heavily used in groups I am involved in or close to:
FFDN, DIY ISPs, Éxégètes Amateurs, Framasoft, ... (all being horrible
leftists and libre-ists). It occurred to me that its bugs,
administration costs and limitations were hampering us.
It realistically aims at replacing Etherpad Lite with something better
on every aspect for both clients and servers: lower CPU, memory and
network usage, more features, fewer bugs and an active development.
The code is hosted on https://gitlab.com/adrien-n/prose and can be
downloaded either through git or through tarballs on
https://gitlab.com/adrien-n/prose/tags .
A demo is available on http://prose.yaxm.org/pads/caml-announce . Any
document name can be used and the website root is an alias for
"default".
Its development has been in line with the vision of the new French
President to make France a « Startup Nation ».
As such, the current release works, has an UI that shouldn't change too
much but also has a few caveats that aren't immediately visible. It is a
« minimum viable product », i.e. « a product with just enough features
to satisfy early customers, and to provide feedback for future
development. » [1]. It is believed the AGPLv3 license will scare
absolutely no angel investor.
I haven't conducted thorough benchmarks because performance is very
clearly in favor of Prose:
- lower network usage for cold and hot browser caches, almost optimal,
- much much lower CPU usage
- > 4 times lower server-side memory
- > 5 times lower page load time
Installation is not documented through an Ansible role which is stored
under ansible/ in the sources. Ansible code is not very fun to write but
reading it shouldn't be an issue and it guarantees the procedure is
always up-to-date.
The project wouldn't exist without Ocsigen and QuillJS.
QuillJS [ https://quilljs.com/ ] is a rich-text editing widget which
features Operational Transform [2] for edition deltas.
Ocsigen [ https://ocsigen.org/ ] is already well-known in the OCaml
world; it does most of the hard work an everything else is based on it.
The code has been quite heavily documented and its entrypoint is
prose.eliom.
One of the project goal has been to make an ocsigen project that could
be used by others to learn. Maybe it has now grown too much for that but
it should still be possible to extract interesting blocks and make a
nice tutorial using it.
Besides general improvements, future version will introduce client-side
storage, session-handling and encryption along with more export formats.
Tests, bug reports and contributions are warmly welcome. Development is
carried on gitlab which handles oauth2 and authentication using accounts
from twitter, facebook, github and bitbucket is possible.
[1] https://en.wikipedia.org/wiki/Minimum_viable_product
[2] https://en.wikipedia.org/wiki/Operational_transformation
Archive: https://sympa.inria.fr/sympa/arc/caml-list/2017-07/msg00027.html
Daniel Bünzli announced:
In order to make it easier to use opam v2.0.0's pre-releases on macOS, there is
now a homebrew-based, quick and foolproof workflow that makes it safe and easy
for you to switch from 1.2.2 to 2.0.0, and back with your previous setup if
needed.
The instructions are available here:
https://github.com/ocaml/homebrew-ocaml/wiki/From-opam-1.2.2-to-2.0.0-and-back-with-the-OCaml-tap
You can now try to live day to day on opam 2.0.0 with the peace of mind of being
able to come back in a breeze to the solid and reliable opam 1.2.2 you know if
you hit any wall.
Archive: https://sympa.inria.fr/sympa/arc/caml-list/2017-07/msg00038.html
Gabriel Scherer announced:
We are happy to announce a hacking session for languages of the ML
family in the evening of July 13th in Pittsburgh, PA, US. The event
will take place at CMU, between 16:30 and 22:30. Access instructions
can be found at:
http://www.ccs.neu.edu/home/gasche/tmp/ml-languages-hacking-session-july-13-2017/announce.html
This event, organized by Anton Bachin, Adrien Guatto, Ram Raghunathan,
Gabriel Scherer and Jon Sterling, is open to all programmers in
a ML-family language, advanced or beginners. Attendees will be
encouraged and assisted in making a contribution to an open source
project, including in particular the OCaml and MLton (SML) compiler
implementations -- we will propose a list of tasks and project ideas,
and try to help in providing technical advice, feedback and guidance
for contribution.
Coming with a project in mind is also welcome. There are other
impactful contribution than code: documentation contributions are
warmly welcome. If you were waiting for an opportunity to scratch an
ML-y itch in a friendly place, there it is!
Event implementation details are still being arranged, but we are not
planning on having food at the event itself. We may go out in groups
around dinner time, but feel free to eat beforehand or bring your own
food.
Apologies for the last-minute announcement, and happy hacking!
Here are some links to messages at http://discuss.ocaml.org that may
be of interest to the readers.
- Calascibetta Romain talks about "Digestif 0.2 - Hashes functions"
https://discuss.ocaml.org/t/ann-digestif-0-2-hashes-functions/504/1
- Lélio Brun talks about "Obelisk 0.1.0: a tool to pretty-print Menhir files"
https://discuss.ocaml.org/t/ann-obelisk-0-1-0-a-tool-to-pretty-print-menhir-files/509/1
Here is a sneak peek at some potential future features of the Ocaml
compiler, discussed by their implementers in these Github Pull Requests.
- stdlib: Added fold, filter, exists, for_all to string/bytes
https://github.com/ocaml/ocaml/pull/882
- Compiler drivers: make -keep-locs the default
https://github.com/ocaml/ocaml/pull/1219
- tweak GCC options to try to work around the Skylake/Kaby lake bug
https://github.com/ocaml/ocaml/pull/1228
- unique names for weak type variables
https://github.com/ocaml/ocaml/pull/1225
- Toplevel: only escapes bytes and not strings
https://github.com/ocaml/ocaml/pull/1231
- Add Unicode character escape \u{H} to OCaml string literals
https://github.com/ocaml/ocaml/pull/1232
Here are links from many OCaml blogs aggregated at OCaml Planet,
http://ocaml.org/community/planet/.
Full Time: Software Developer (Functional Programming) at Jane Street in New York, NY; London, UK; Hong Kong
http://jobs.github.com/positions/0a9333c4-71da-11e0-9ac7-692793c00b45
Provisioning, deploying, and managing virtual machines
https://hannes.nqsb.io/Posts/VMM
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.