ITEC 325 2015spring ibarland

home—lects—exams—hws
D2L—breeze (snow day)

lect42-wrap-up
retrospective

What did we talk about this semester, eh?

• client-side vs server-side.
  That is, being able to figure out answers to questions like:
  • Where will code for "add to database" be?
  • How about code for "de-select checkbox if textbox is empty" be?
  • How about code for “Once the state dropdown is selected, auto-fill another”?
  • What choices does browser have? The server?
  • How many requests are actually made?
  • What is some of the info that must be in the http header for requests, and for responses?
• php basics: functions, associative arrays (hashes), iteration, etc.

  Note: php is pass-by-value. So if I call $_POST = tidyArray( $_POST ); how many times is the array being copied?
  A: four! (to & from function; once by me manually in loop; once by the '=' !)

  • 4x copying an array of 20 items isn't going to be bad
  • 4x copying large arrays when I'm trying to serve 100pages/sec? That's more serious! Pass-by-ref.
  • pass-by-ref: but beware; tends to lead to hard-to-track bugs
  • why not just avoid passing altogether, and use $_POST (super)globally? Many will suggest that, but:
    • if you switch to GET then you'll need to write (and maintain) another copy of that function;
    • if tidyArray is a general purpose function, you'll *want* it for all arrays.
  • Handy idioms: array_map (and, array_walk), where you pass in a function [well, a string — the function's name!]

• passing state between pages (since http is itself stateless)
  forms (including hidden-inputs); cookies; files; sessions; database
• XML: hierarchical data (tree) marshaled to text format; human-readable; having a standard is great.
• XSLT: a langauge for XML -> XML functions.
  • it is itself XML
  • processed client-side (but, server-side libraries also exist)
  • uses xpath expressions (in xsl:foreach, xsl:value-of, xsl:choice etc):
    • xpath: expressions evaluate to a *nodeset*.
    • absolute vs relative paths -- relative to the node currently being handled (by template, or in loop)
    • filter by using square brackets: nodeset-expr[boolean-expr]

What's not to like?

How cool is php? Nice, low-barrier entry But it makes me feel kinda like I'm using assembly-language: To make a site about kittens, I need to...

• have a web form, w/ fields for kitten's name, vaccination date, #claws, cuteness level, ...
• do javascript validation on each of those fields
• print out a kitten's info as html
• have php that validates each of those fields
• (maybe: php objects)
• define a database table for kittens
• set up SQL types/constraints for each of those fields
• (maybe also: a java program that interacts with the database: class Kitten { ... } and the constructor with all its validation code for sanity-checking.)

Do we really need a different language for each different stage?

Many before have seen this problem, and sought for a better world.

• JQuery: These are higher-level functions that do common tasks that we're coming to expect (incl. nice visual effects of I/O)
• Node.js: a library; meant for running .js both client and server-side.
• More comprehensive approaches like: .NET, .asp, rails -- specify what info a Kitten has just once, and let the platform will auto-generate html forms, client-side checking, server-side validation, database def'ns, a java class, etc..
• Another example: c# data annotations.

So the php/javascript in this class grounds you in the 'assembly-level' web programming: you now know/appreciate the low-level work that needs to be done. But don't think that just because most sites still use just php/javascript, that it's web programming's highest level!

home—lects—exams—hws
D2L—breeze (snow day)

©2015, Ian Barland, Radford University Last modified 2015.Apr.27 (Mon)

Please mail any suggestions (incl. typos, broken links) to ibarlandradford.edu