Nonstandard Testing: Fill & Download for Free

Comment thoroughly, why it’s needed and what it does and how all variables are usedReuse modules as often as possibleAvoid use of a callable command structure where you pass in the name of the called procedure; it's often horribly complex if not unwieldy to test.Don't change the tests module when testing subroutinesSet all variables to zeroDon't plagiarize, but do reuse, keep the original programmers name on the moduleTest the user interface first with the customer and make it as easy as possible to use, keep it separate from other codeSeparate out nonstandard OS, database or other hardware specific code and flag it for essential testing at every blockpoint (avoid this use as much as possible)Don't nest looping constructs without an exitHave someone else do it (I don't do any anymore, no need, I'm 60)

No."A standardized test is a test that is administered and scored in a consistent, or "standard", manner."- http://en.wikipedia.org/wiki/Standardized_testThe value of standardized testing is not in education, but in credentialling. A credential has more value when more people understand what it means- using a nonstandardized test automatically means that the test is less valuable to people who are not experts in the field.Consider a hypothetical situation. You are a budding Picasso who wants to pursue your art. However, you are from a disadvantaged situation and can't afford to attend art school. It is very difficult to prove that you are worthy of scholarship, or even admission, among peers who can spend more time and effort on grooming and traveling, applications and honing your art, because there is a distinct lack of a standardized test for artistic ability. This could be for many reasons, but for the purpose of our hypothetical the fact is enough.Contrast this with a math or physics prodigy- due to these fields having clear standards and testing mechanisms, you will be identified as a standout extremely early in life, and you will be eligible for many scholarships and special education programs based on your test scores. It will even be easier to convince potential sponsors to be your patrons, because a larger pool of patrons will be able to compare you more quickly with age-peers and see how outstanding you are.The short version of this argument is- abandon standardized testing in favor of what, and how could a lack of standardized tests benefit students, considering that the quantification of students through standardized tests is the mechanism by which excellence can be recognized, as well as weakness improved, in fields regarded as valuable by the entities which fund schooling.However, I understand the spirit of your question to be- WTF is wrong with standardized testing, it's destroying our schools because kids are (1) only learning things that fit into existing standardized frameworks, and then (2) only to metrics set as averages across the entire population, which (3) screws the excellent students and encourages mediocrity in the rest!Lets look at how improved and expanded standardized testing could actually help solve all these problems.1. Standardized Testing requires that students conform to existing standardized fields and frameworks!Education could expand standardized testing into new fields (art, music, literature, english composition, etc), understanding that each student's unique makeup is worth specialized support, and encouraging each student's particular excellences while encouraging only mediocrity in 'core skills.'Standardized testing can easily be applied to fine arts, but honestly no one has really tried, except in historical aspects. This is due to no one wanting to stand out and say 'this is art, nothing else is art. If you do those other things you fail at art.' Art, as the expression of creativity, is widely considered to be incomprehensible, yes, the direct opposite of standardized testing. However, the information collected by standardized tests gives a range of normal answers- it confers equal information to get a 0% on a 1,000,000 question test as it does to get 100%- with either one you need to know all of the correct answers to overcome the margins of error and achieve your standout score.Modern statistics actually doesn't have a problem identifying _really weird_ testers. If good art is done by really odd people, you can identify that just as easily with standardized testing as you can identify perfect performers. Similar statistical measures can be applied to writing (parse the entire document, is your word count normal? What % of your words qualify as misspellings? If writing poetry, do you properly conform to the limits of the form?) and once grammatical competency is established, compare it to the person's weirdness, and you can guess that with enough resources someone who can do perfect form and be perfectly not-normal is probably worth teaching how to paint.This applies in other ways too. Not every student has the ability to study random words and numbers as much as possible to max out a score- but you can easily build a test which generates random words and letters for each student, give them a period of time, and see which ones get the highest retention scores.Here's an interesting fact about that. Imagine you've got students who are only learning _completely randomized and useless information_ like in the prior example, their whole lives, until they can't take it anymore and quit. It confers a lot of information about the student to see what their breaking point is. Personally I like hiring people who break fast- they don't put up with pointlessness. http://tvtropes.org/pmwiki/pmwiki.php/Main/YouWereTryingTooHardIn short- standardized testing doesn't require that students conform to existing standardized fields and frameworks, but the incentives and policymakers who use standardized testing in historically constrained manners.2. Kids are only learning to metrics set as averages across the entire population!The controversial No Child Left Behind act had many consequences- the most relevant here is that it united everyone who had children going through schools hamstrung by it in hatred of it.I'll be more brief in addressing this one- there is no need to average children and then use all available funding to improve the worst performers, while ignoring the best. This is failed standardization at its worst, promoting mediocrity at the expense of excellence. Attempting to ensure that all educated children reach a mediocre benchmark may, perhaps at worst, result in success. How to fix it without removing standardized testing? Simple- repealment of the act, or expansion of it to include the above metrics for creativity and 'weirdness.' Either would do a great job of addressing the issue.3. Excellent students are being screwed, and only me mediocrity is encouraged in the rest!This perspective actually has a built-in fallacy- that all students are either excellent or mediocre, at best. A much wider set of standardized tests, including in each the potential for weirdness using the above techniques, shows the wide variety of skill and aptitude in any tested population.My suggestion for this is that students who perform well on weirdness in any category be encouraged into that, and that students who perform well in normal fashions be encouraged in those areas, via less structured classwork in which every teacher teaches those who come into their classroom in a collegiate fashion- basic lectures, advanced students assisting lower students in performing on area-specific standardized tests, and advanced students also pursuing their own studies more directly under the teacher. Expansion of possible fields of study to include 'weirdness' in classically defined fields of standardized testing would do much to ensure excellence is rewarded in fields apart from specialties, encourage a mediocre level of understanding in all students, and allow students doing poorly to actively subvert the fields of difficulty- a historically attractive choice for students of excellence and mediocrity alike.

Lots and lots of them:Database access bugs, particularly database performance issues. The query with the gnarly join may appear to work in your tiny developer db, but (hopefully) your integration db is big enough to catch your lack of a crucial index, etc.All manner of networking, protocol, or file-format bugs. Most people don't bother to test code dealing with this type of stuff at the unit level.Bugs related to interactions with peripheral devices.For apps with Web interfaces, bugs related to browser incompatibilities. (The bug may be because the browser has a nonstandard implementation of some function, but if it breaks your website, it's your bug.)Bugs related to third-party libraries or frameworks. Most people don't unit-test these interfaces.Having once worn a QA manager hat, I prefer a large body of diverse integration tests - and easy ways for developers to run automated integration tests - over lots of unit tests. Unit tests often have the problem of "looking for the keys under the streetlight" in that they test what the developer thinks can break versus what actually does break.