The Top Twelve Open Source Projects of 2017

2017 has been a year in which open source soft­ware has flour­ished. This pre­sen­ta­tion of­fers a bird’s eye view of the past year’s best of­fer­ings in soft­ware.

OpenSource For You - - Contents - By: Swap­neel Mehta The au­thor has worked with Mi­crosoft Re­search, CERN and star­tups in AI and cy­ber se­cu­rity. An open source en­thu­si­ast, he en­joys spend­ing his time or­gan­is­ing soft­ware de­vel­op­ment work­shops for school and col­lege stu­dents. You can con­nect

One of the world’s most re­puted tech­nol­ogy re­search firms, Gart­ner, had pre­dicted early in 2017, that the next decade would be spurred by a whirl­wind of ad­vances un­der­girded by a fo­cus on ‘ar­ti­fi­cial in­tel­li­gence ev­ery­where’, ‘dig­i­tal plat­forms’ and ‘trans­parency’. Lit­tle did we know that the pre­dic­tion would ring true from the fol­low­ing month it­self.

It has been a land­mark year for open source! We are on the cusp of a revo­lu­tion, poised at the fron­tiers of the tech­no­log­i­cal re­birth of our world, trig­gered by a spec­trum of pos­si­bil­i­ties that are open­ing—from 5G to blockchain – all of which are in­flu­enc­ing busi­nesses to adopt a de­cen­tralised ap­proach in an ef­fort to keep up with the times.

With the game chang­ing re­sults pro­vided by deep learn­ing al­go­rithms, the tech in­dus­try is pay­ing close at­ten­tion to all the pos­si­ble use cases. From Google and Uber to Baidu and Tesla, al­most ev­ery ma­jor tech com­pany has in­vested heav­ily in ap­ply­ing such prac­tices in a bid to im­prove their of­fer­ings and in­tro­duce novel fea­tures for the con­sumer.

The re­cent hype also sent cryp­tocur­rency prices soar­ing, be­fore th­ese are plum­meted back to rea­son­able lev­els. The 2017 Christ­mas sea­son was pep­pered with ini­tial coin of­fer­ings, as en­trepreneurs went all out to har­ness pos­i­tive pub­lic sen­ti­ments and chan­nel th­ese into VC fund­ing. In fact, the largest play­ers in the fi­nan­cial mar­ket have set up con­sor­tiums to look at how th­ese tech­nolo­gies can be in­cor­po­rated into their busi­ness mod­els in the cur­rent cli­mate. All this could lead to rad­i­cal changes in the rather ar­chaic and con­strained in­sti­tu­tions that have driven the fis­cal as­pects of our lives thus far.

So, all in all, it’s been a highly tran­si­tive year as we move into a decade that will res­onate with per­va­sive ar­ti­fi­cial gen­eral in­tel­li­gence. We need to come to terms with a huge leap from the sta­tus quo into bleed­ing edge tech­nol­ogy.

Top open source projects of 2017

A great year for open source projects, 2017 wit­nessed a num­ber of ma­jor changes to ex­ist­ing projects and an as­sort­ment of new projects tar­geted at all as­pects of the soft­ware de­vel­op­ment life cy­cle. Some of the most pop­u­lar of­fer­ings have been high­lighted in this ar­ti­cle. The order is

in no man­ner rep­re­sen­ta­tive of rel­a­tive im­por­tance.

1. Hy­per: A ter­mi­nal built us­ing HTML, CSS and JS

The cre­ators of Hy­per had one ba­sic goal in mind— to cre­ate a sim­ple, ef­fi­cient and, most im­por­tantly, hack­able com­mand line in­ter­face for de­vel­op­ers. In­tu­itive short­cuts can be added in Hy­per based on de­vel­oper pref­er­ences, and there is the op­tion of build­ing your own plug­ins to pro­vide ad­di­tional func­tion­al­ity to this tiny but pow­er­ful ap­pli­ca­tion.

2. Parse Server: The open source ver­sion of the Parse back-end

This soft­ware can be de­ployed in any in­fra­struc­ture that is ca­pa­ble of run­ning Node.js, and works well with the Ex­press Web Ap­pli­ca­tion Frame­work. It can be de­ployed in­de­pen­dently, or added to ex­ist­ing Web ap­pli­ca­tions. Sup­ported by an ar­ray of tu­to­ri­als and ex­ten­sive doc­u­men­ta­tion, this has proved to be a pop­u­lar, com­mu­ni­ty­backed project.

3. Ten­sorFlow: Google’s open source ma­chine learn­ing frame­work

Orig­i­nally re­leased in 2015, the Ten­sorFlow project saw ma­jor up­dates in 2017. Th­ese in­cluded sup­port for Python gen­er­a­tors, ad­di­tions to var­i­ous APIs for per­for­mance im­prove­ments and, most im­por­tantly, the ad­di­tion of Keras to the Ten­sorFlow core pack­age. In ad­di­tion, Ten­sor­board was re­leased to help im­prove vi­su­al­i­sa­tion via plot­ting of quan­ti­ta­tive met­rics on a graph. Ten­sorFlow is cur­rently the most pop­u­lar project on GitHub with over 77,000 stars. In com­par­i­son, the Linux project has just about 53,000 stars!

4. Caffe2 and PyTorch: Up­dates to both open source ma­chine learn­ing frame­works

Face­book-backed PyTorch was de­vel­oped in re­sponse to Google’s re­lease of Ten­sorFlow. Face­book also uses

Caffe for train­ing deep learn­ing mod­els in pro­duc­tion. An ex­tremely in­ter­est­ing de­vel­op­ment in the re­cent re­leases of both th­ese frame­works is due to Face­book and Mi­crosoft’s part­ner­ship on re­leas­ing up­dates to th­ese frame­works based on a new pro­to­col de­fined as Open Neu­ral Net­work Ex­change (ONNX). ONNX is a stan­dard for the rep­re­sen­ta­tion of deep learn­ing mod­els that al­lows mod­els to be trans­ferred be­tween frame­works. It is go­ing to be in­ter­est­ing to mon­i­tor progress on this front.

5. Bulma: A FOSS CSS frame­work based on Flexbox

Avail­able as a more in­tu­itive al­ter­na­tive to pop­u­larly used Boot­strap, Bulma is es­sen­tially a sin­gle file— bulma.css; it does not in­clude any JavaScript since peo­ple gen­er­ally want to use their own JavaScript im­ple­men­ta­tion. It can be con­sid­ered ‘en­vi­ron­ment ag­nos­tic’ since it’s just the style layer on top of the logic.

6. Anime: A light­weight, open source JS li­brary tai­lored for an­i­ma­tion

This is a light­weight JavaScript an­i­ma­tion li­brary. Anime works with any of CSS’ prop­er­ties, in­di­vid­ual CSS trans­forms, SVG or any DOM at­tributes, as well as with JavaScript Ob­jects.

7. Yarn: A fast, re­li­able and se­cure de­pen­dency man­age­ment soft­ware

In­tro­duc­ing a num­ber of un­con­ven­tional and in­no­va­tive fea­tures, Yarn presents a con­ve­nient al­ter­na­tive to the soft­ware in use. What it of­fers ranges from off­line in­stalls on ac­count of caching, to in­creased re­li­a­bil­ity and se­cu­rity, aris­ing from par­a­digms within the in­stal­la­tion that of­fer net­work re­silience and con­cur­rency for per­for­mance.

Yarn goes the dis­tance for de­vel­op­ers look­ing for more ef­fi­ciency and less fail­ures across sys­tems re­sult­ing from mis­man­age­ment of de­pen­den­cies.

8. Apache Soft­ware: Mul­ti­ple projects by the Apache Soft­ware Foun­da­tion

From re­leases of Hadoop to Spark, the Apache Soft­ware Foun­da­tion has had a num­ber of im­por­tant re­leases for projects in the open source do­main this year.

Car­bonData is a novel Big Data file for­mat for in­ter­ac­tive query­ing us­ing ad­vanced colum­nar stor­age. TinkerPop is another project that works be­hind the scenes in pow­er­ing graph-style mod­el­ling, for­mu­la­tion and anal­y­sis for a num­ber

of pop­u­lar frame­works in­clud­ing the likes of Spark, Ti­tan and Neo4j. Kudu, an in­te­gral com­po­nent of most en­ter­prise stacks al­ready, is poised to re­de­fine the way we ap­proached data stor­age and anal­y­sis us­ing Hadoop and HBase, pro­vid­ing an op­ti­mal so­lu­tion for large amounts of fre­quently up­dated data.

MXNet was one of Ama­zon’s widely dis­cussed deep learn­ing frame­works for cross-plat­form ap­pli­ca­tions on the Web and the mo­bile, that was ac­cepted by the Apache In­cu­ba­tor and saw ex­cit­ing progress through 2017.

Some other projects with ma­jor up­dates re­leased in 2017 by the Apache Soft­ware Foun­da­tion in­clude Zep­pelin,

Solo, Ar­row, Spark and Kafka.

9. Cock­roachDB: A mas­sively scal­able, fault-tol­er­ant SQL data­base

As the name states, the in­spi­ra­tion for this project is from cock­roaches, the only liv­ing crea­tures that sci­en­tists be­lieve will sur­vive a po­ten­tial nu­clear war or ice age— the data­base is mod­elled sim­i­larly to of­fer an SQL in­ter­face that can scale mas­sively, re­li­ably, and fo­cus on re­silience in­stead of re­cov­ery. Cock­roachDB is serv­ing global cloud needs for or­gan­i­sa­tions like Baidu.

10. Ku­ber­netes: Con­tainer or­ches­tra­tion and man­age­ment at scale

For soft­ware that was adopted by a mere 10 per cent of the in­dus­try in 2015, to rise to a phe­nom­e­nal adop­tion rate of nearly 71 per cent of the mar­ket by 2017 (ac­cord­ing to the same ser­vice provider) is an achieve­ment. Orig­i­nally a Google-backed project and now man­aged by the Cloud Na­tive Com­put­ing Foun­da­tion, Ku­ber­netes is only go­ing to see good times as cloud com­put­ing and con­tain­ers take the tech in­dus­try by storm.

11. XGBoost and CatBoost: Gra­di­ent boost­ing li­braries for ma­chine learn­ing

The ad­vent of ma­chine learn­ing in 2017 saw the fo­cus shift to the prac­tice of gra­di­ent boost­ing, of­ten touted as ‘steroids for learn­ing al­go­rithms’. While XGBoost works on multi-lan­guage and multi-plat­form gra­di­ent boost­ing, CatBoost of­fers cat­e­gor­i­cal fea­ture sup­port for gra­di­ent boost­ing on de­ci­sion trees.

12. Mi­crosoft .NET Core 2.0: A min­i­mal sub­set of the .NET frame­work

Placed at the fore­front of Mi­crosoft’s foray into open source, .NET Core 2.0 of­fers im­proved func­tion­al­ity to de­velop cross­plat­form ap­pli­ca­tions for the Uni­ver­sal Win­dows Plat­form as well as Xa­marin. It will be in­ter­est­ing to see how adop­tion rates shape up for this project in the days to come.

Fig­ure 1: 2016-2017 trends in open source adop­tion (Im­age courtesy: Rackspace)

Fig­ure 2: The top open source projects in 2017

Newspapers in English

Newspapers from India

© PressReader. All rights reserved.