Irrational fear of scriplets - revisited

Question

hey all this is a continuation of a very long thread that begins here
http://forum.java.sun.com/thread.jsp?forum=45&thread=318105. (if you read it, remember to post any continuations on this thread.
My basic contention is that small scriptlets which do things like get values from beans, or do simple conditionals are fine and that forcing people to use the tag libraries which do these things is not productive. For the record, I advocate MVC, I advocate good architecture, But the debate about tags vs WELL-DESIGNED scriplets usage (i.e. a line or two which call a bean or regular java class, or just does a loop over some collection)is not about architecture it about syntax, and the ethereal concept of readability. (And fear.)
Why do they exist then?
Managers want to make the sofware development process faster and cheaper. Architects somewhere along the line have sold them the notion that a syntax that has carets in it is somehow easier to learn (dumb manager says, "hey that looks like HTML, must be just as easy", sleazy tag salesman says,"oh yeah, it is. Your dog could do web apps now"). The management team then thinks about the savings they could reap with pound puppies. The big flaw in all this is that these tags are not easier to understand than the scriplet which would do the same thing. Why invest time in learning this library of arcane Collection management tags, when a scriplet would just be standard java. Standard Java is here to stay, the Expression Language (EL) and some of the more pointless libraries are a little more "iffy", and consequently not as big an asset to ones' skill set.
Detractors of this argument would say that these tags are intended for "nonprogrammers" or "GUI Developers", but, I'm sorry, once your dealing with hierarchies of objects with the collections tags, your programming. Like it or not. So why not use a real programming language. When james gosling invented java why didn't he do the whole thing with tags, if they're so readable? Also remember that any decent html developer, also knows javascript. How far really is the leap from javascript, to the small amount of java required to call an java class from a scriplet? One of my good hearted compatriots in the original thread suggested that The leap is too far because java is a typed language and javascript is not. Come on, that just doesn't fly. So they should read the chapter on objects, types, and casting. And then come in the next day and get to work. Whatever mental facility would need to be in place to learn the small amount of java to access a class in a scriptlet also must be in place to learn EL and collections tags. Now they're becoming java programmers and maybe architects feel they must continually reinforce the notion in the minds of management that there is an unbending hierarchy of programmers and that some people could just never do certain things. This attitude is necessary to maintain the hierarchy at the top of which the Architect presides. I guess from the management perspective, I would be upset if I taught all my html developers, java and then they all wanted more money cause now they're java developers. But then again, from the human being perspective I don't like to see the careers of others actively stunted. And besides management should have the foresight to see the possible glut of new java programmers would force rates down. It's a win-win-lose, the loser is the Architect whose rate could drop due to more java programmers being around. But he was already a loser. This argument is almost a moot point. Every project I've been on has java developers doing the pages anyway.
I realize that in spite of all these arguments, architects will still maintain the status quo, categorizing the potential of other people into hierarchies of their liking. It's like dealing with Liberals. The main argument always seems to return to readability assertion which is rarely ever challenged, except in private by those who don't want to appear "stupid" by challenging the orthodoxy of their industry. I, However have no problem appearing stupid. And do it quite well, sometimes. Maybe I'm doing it now. I do allow that I could be totally wrong. Someone convince me.
I always held the Unix, Linux, Java world in high regard because they refused to participate in the "Dumb it Down", "Hide The Details" approach that their major rival sells. (Tags are not really even a dumb down in my opinion).
And the parting question. Is there a way of objectively determining readability? Who determines readability?
p.s. check out this thread about a developers frustration with EL and then think about how much easier the scriplet solution would be.
http://forum.java.sun.com/thread.jsp?forum=45&thread=304696

Answer 1

Hi once again jfoltz2.
Trying to raise this topic again I see. Good on you. As you know (from my postings on the original thread), I very much sympathise with you regarding the 'irrational fear of scriptlets' that seems to have infected a substantial number of JSP developers.
It was a pity the original thread died, because while there were some quite strongly held beliefs that TAGLIBS were the bees knees. By the end there were a number of others expressing caution about blind adoption TAGLIBS (mostly from smaller teams/organizations I noticed). It's not so much that TAGLIBS are bad per se, but that they are not a panacea. I particularly enjoyed jlusk4's posting (the final one) which summarised many of my views and concerns too (hope you rediscover the thread jlusk4).
The original thread also strongly re-enforced concerns I have about just how poorly integrated our web tools/languages are. While the details may be different for each user and application, I don't think it would be uncommon for developers of interactive web sites to use Java/JSP/TAGLIBS/XML/Databases/etc on the Server side and HTML/Javascript/DHTML/CSS/Flash/Java/etc on the Client side. Each with it's own quirks and limitations. They are far from seamlessly integrated.
When you consider all the different browsers, browser revisions, differences in implementation between core browsers technologies (e.g. Javascript), container engines (e.g. tomcat), etc. It is clear that developing interactive web sites is difficult and not without pain, in fact it's rather surprising that it works at all. The multiplicity of languages and tools is, in my view, unnecessarily complex and cumbersome and breeds a culture of gurus and wizards (which isn't very helpful either). In short it wastes time, energy and money.
Okay I've gone slightly off topic, so I'll try and bring it back on track. Clearly scriptlets don't help reduce the complexity of integration, but TAGLIBS don't add anything magical either. There simply represent an alternative band-aid/plaster to a more fundamental problem. We should keep this in mind when we're debating whether scriptlets vs TAGLIBS.
Finally, from a sociological point of view, the original thread was interesting, because it demonstrated just how strongly people hold their beliefs about software design (myself included). Holding strong opinions is fine so long as you understand the others point of view and are willing to accept that you might be wrong. From that basis you can make a balanced judgement, for YOUR own work area. Being a zealot helps no one, not even yourself.
Having said that I have to confess that I am a zealot in one way. I'm intolerant of the intolerant.
Anyone want to take the bait ?
Cheers,
Jason.