Komentar #646265

Još prije nešto više od dvije godine komentirao sam post na jednom SitePointovom blogu, pod naslovom 10 Years of Java… for what? SitePoint mi od tada automatski šalje obavijesti o svim komentarima koji su postani nakon mog, i logično je da su se s vremenom ti komentari prorijedili i naposlijetku sasvim prestali.

Zato me je prilično iznenadio jučerašnji mail kojim sam obaviješten o novom komentaru, dodanom dakle pune dvije godine nakon diskusije. Obično takve komentare letimice pročitam i odmah obrišem mail, ali ovaj me je iznenadio prije svega svojom dužinom, ali i sadržajem, zbog čega ga prenosim u cijelosti i bez ikakvog komentara.

I samo ime kojim se potpisao autor komentara samo po sebi dosta govori: paul works with java everyday at work since ‘94

hi-

I have been working with j2ee in corp. shops in the usa (boston) since nineties. I agree wholeheartedly with the original poster.

It is possible to see j2ee corp apps that run fast, but the entire j2ee tech stack must be configured perfectly and every line of java code perfectly written. Never happens in practice, or very rarely.

Programmers that write in java are less productive because of all the framework configuration, discovery and learning curves, etc. that are needed to build the scalable applications they are tasked to create. I work on j2ee apps that scale, but badly - poor performance and low availability at high numbers of concurrent users.

RIA with applets is a no-go, and a j2ee programming team capable of RIA without applets in j2ee along the lines of the very impressive thinkfree office will not be easy to assemble. applets in general are a deadend and the incompatibilities are astounding on a WORA-billed platform.

The entire universe of j2ee accoutremont: maven, ant, hibernate, jrun, junit, jmeter, on and on… the xml configuration files, all of this adds complexity, and team members, and increases code and chances for misconfiguration and errors in deployment.

Performance of j2ee is much better in recent years, but in 2008, i still use s-l-o-w j2ee applications every single day. the author of the post is right: ten years with little to show for it.

j2ee means all to frequently: no firefox ’cause we can’t support it, no mac, no linux. so, j2ee in the corp world means windows and IE. Not very hackerish. Or cool. Or fast. And I love windows as much as the others. I’m knocking most corp j2ee app development in the us, certainly 95% of it in boston in 2008.

The j2ee web application servers, and servlet containers frequently have their own problems which delay deployments and crash sites, etc. Things like weblogic, tomcat. Sometime you work in a j2ee shop that has one piece of the j2ee chain that’s a little older and it holds things up during deployment, retards performance of the j2ee application, or causes a recurring availability issue.

Jboss and related and similar j2ee enabling and complementary technologies are developed by toolmakers and when added to all the frameworks out there (ICEfaces, millions more), before you know it you’re using tech from dozens of vendors and keeping track of the lifecycles and updates becomes unmangeable. How about five, ten years from now? Where are the j2ee apps then? Who maintains them?

j2ee programmers need to know so many j2ee technologies that they even the best ones lack experience across the entire spectrum of technologies and so they learn on the job. And that takes time, and makes projects using j2ee later. They have less time to learn about anything but j2ee, and they think (my personal observation of many of them) you don’t know anything about computers if you don’t know java. Many have no experience of anything before java.

I already said above, j2ee apps scale - mostly not that well. Again, in my personal experience.

The JRE is not great. It is the weak link in many j2ee apps. Frequently incompat. between different j2ee apps which need a diff. version.

I have never ever seen a j2ee dev team that did anything more than pay cursory lipservice to agile development methodology.

I have seen during the nineties ecommerce buildout many scaled perl and scripting language LAMP applications and sites that filled millions of orders, charged millions of dollars, served millions of pages. So I know they scaled, because I saw them. J2ee’s scalability, to my mind, is frustratingly unevident when judged against these personal recollections. Most of the j2ee applications i have worked on slow down the more people are using them, etc. This is based on my daily working life in many j2ee-only shops in boston. I work with it for a living, and j2ee has paid my bills and put food on my table. I’m just simply agreeing with the original poster’s comments.

j2ee UI’s and UI development (JSF, java server faces): ack. Swing, struts, AWT: it just never happened correctly. Every j2ee dev team i work on has poor UI. I know, I know, your experience is different. Mine has been - java, j2ee UI is usually bad. I’m working on a VUI dialog designer for work that is an applet that interfaces with a j2ee backend. The UI is good, almost very good. I mentioned thinkfree office above. But almost all j2ee apps have poor UI and interface. In ten years, j2ee UI will be one of the main complaints about j2ee in hindsight, along with bloat, complexity, gross overuse, inappropriateness for a wide variety of programming problems, perfomance. Java is still too slow too often in 2008.

Gotta go work with more j2ee tomorrow at work, so I have to go. Just kidding about java and j2ee. It’s great.

Popularity: 26% [?]

1 Comment so far

  1. vojimir golem on March 5th, 2008

    Česta je zabuna da se J2EE komparira sa drugim “web frameworks” sa aspekta produktivnosti, a onda se takav “zakljucak” prenosi na Java platformu kao cjelinu.
    Ako govorimo o EE onda da ne bi trebali upadati u “kruske i jabuke” antipatern te bi trebalo vidjeti sta stvarno upada u istu kosaru.

    Ako zelimo uporedjivati npr RoR onda je pandan npr Grails. Ako govorimo o JEE onda je “konkurencija” CORBA nekad ili .NET danas.
    Ako govorimo o “Ajax toolkits” onda je “konkurencija” mozda GWT.

    Nikako ne zelim upasti u zamku tvrdeci da se na LAMP ili RoR ne mogu raditi aplikacije i za enterprise ali to nije ista tvrdnja.
    U automagazinima rijetko vidjam “usporedbu” novog Mercedesovg kamiona i Porcheovog sportasa, u IT-u nazalost dosta cesto.

    Ako govorimo o inovativnosti pogledajmo malo utjecaj na MS: Java - C#, J2EE - .Net, GWT - Volta.
    Kvaliteta neke tehnologije generalno najbolje se vidi kada i konkurenti počnu preuzimati ista rješenja.
    Utjecaj npr. Eclipse-plugin modela na Visual Studio tek se ocekuje u punoj mjeri.

    Daleko od toga da svi oni nisu utjecali i na Java platformu. Dobra praksa i sa LAMPa i sa .Net ulazi i u Java ekosistem.

    Da Java ekosistem ne pati ni od “not invented here” antipaterna dokaz je i Grails/Groovy.

    Svaka tehnologija ima svoj zivotni ciklus, pa ce sigurno doci i vrijeme zalaska Jave i kao jezika i kao platforme, iz jednostavnog razloga sta samo prilagodjavnje nece biti dovoljno vec ce biti potrebno razviti platformu na nekoj novoj - bitno razlicitoj paradigmi.

    U 2006/07 cinilo se da su dinamicki jezici mozda taj preokret, ali izgleda da nije tako.

    Po mom misljenju 2008 je upravo godina kad je uocljiva “renesansa” inovacija bas baziranih na Java jeziku, platformi ili JVM-u. Samo “Google Android” je za mnoge nove developere dovoljan razlog za ucenje Jave kao jezika.

    To sto se autor komentara ne snalazi u Java ekosustavu jest problem (najvise njegov) ali implicitno izvoditi zakljucak da su se Sun, Oracle, Google, IBM i veliki dio OpenSource zajednice opredijelili za neproduktivnu platformu je zakljucak koji vise govori o povrsnom pristupu temi nego li o temi samoj. Da li je kljucni argument sto se covjek bavi Javom od ‘94 pa bi zato trebao biti ekspert na tom podrucju i time zasluzuje da ga se prenese u cjelini?

    Da ne bi bio krivo shvacen ja apsolutno ne sugeriram superiornost bilo kojeg jezika ili platforme. Svi oni imaju svoje dobre i lose strane i njihov medjusobni utjecaj je evolucijski proces.
    U ovom komentaru sam advokat Jave iskljucivo iz razloga sto na blogu autora ciji doprinos IT zajednici cjenim ovom postu, po mom misljenju, nije bilo mjesto.

Leave a reply