Maintainability of an automated test kit - 6

zhaozj2021-02-16  66

The value for automated test is uncertain, such as random), although we need to determine how the test case is used. (Unanimously pass)

We are not sure blind test. Need to know what testing is running, sometimes you need to enter strict and sort of input. But if you decide whether the program is running test, you have to replace those test cases that have been run successfully with a large number of test examples.

We need to design a record of testing the ability to run the log. (Unanimously pass)

Some test tools make record test projects simple, and some become complex. Debug tracking test procedure, you can quickly see the status and return value of the test case.

Human and management

The efforts made by automation work are in version N, which will get more benefits when the version N 1 is released. There is an objection to this authenticity when you can reach a short-term playback target. In the example, it includes smoke testing, pressure test (some pressure tests can only be implemented with automation), configuration / compatibility testing. (Unanimously pass)

If version N is the version of the automation start test, then your basic goal is to write a stable automation script in version N 1. Your second goal is to easily implement except for testing in version N as a goal.

People need to pay attention to automation instead of she as a secondary goal. If no one is put into automation, then automation may be a waste of time. (Unanimously pass)

Many testers have no programming experience, they don't know how to build and design the framework. (Unanimously pass)

Data drive method

This article mainly describes the data driver. I think I'm talking about it is insurance, but there are no people in our imaginary situations. Additional part, detailed remarks in conference.

The topic of the data driver policy contains (for example):

Input program parameters

Execute the operation sequence and command of the program

Test case sequence for driver running

Driver running machine status order

Document recorder reads and operations

Specify parameters and events (such as status models or model-based charts) by the system model (consistent)

Data drive maintenance is small and it is easy to use for non-programmers. (Unanimously pass)

Although we agree with these principles, we have messy or simple thinking examples, and the like. If you lack design work ability, no one will agree and maintain you.

A plurality of interfaces are provided for inputting data to the file for data driver testing. You may choose to provide different interfaces for testers with different experiences. (Unanimously pass)

Frame drive method

Temporarily, the framework discussion enters an expanded discussion of a language design, using the program language is the best practice. We have been trying to modify our list of protocols before others encounter such problems. Retreat along this line we skip the reservation point. Lower is some detailed framework recommendations:

Agree to rely on a certain number of mixers to develop frameworks. (Unanimously agree)

When you create a framework, what level you build is at the level. For example, we must think about three levels:

Menu / Command Level, perform simple commands

Object hierarchy, special processing action to be executed

Task level, pay attention to special, ordinary, repetitive tasks

You may find that he played on a hierarchy, when you really need them, add other use cases.

There are many ways to define and separate hierarchies. Analyze the task with an appropriate method. Avoid unwise to create a very simple test script to test some very up and complex test cases. (Unanimously agree)

The check code of the error handler must contain an error handling when loading the framework function library in the script. (Unanimously agree)

An experience that applies to all developments, especially for test code, because we expect the procedures that are being tested, and to make a sign of failure in order to make reports and processing simple. When you create a shared library command, there is a risk in handling your personal programming and document style. If the developer can't adapt to this, then they will not use the code of others. (Unanimously agree)

Creating a script using your library can save "time". Conversely if you don't create a library. (Unanimously agree)

These libraries are the warehouse for the orderless storage shared command. If a function is very common and requires a large number of parameters, some developers (testers using test tools) would rather use their defined parameters. Some developers do not check what is in the library. Some of the programmers do not believe in the code in the library, because they think that there is (possibly correct) many unspected and errors.

In the data file, the test parameters are required, such as INI files, setting files better than adding constants into the automation script or adds to a file containing scripts. (Unanimously agree)

Package is right. And use this method as usual. (Unanimously agreed)

Localization

We spend a lot of time to discuss localization problems, we have gone a surprising conclusion. We may discuss these issues at the later Lawst meeting, but for automatic localized test experience - If you are told when you pay a lot to invest in GUI, we will have a return, which is a warning for you. These are frustrated representations.

Automated local test goals are the baseline that is still working in previously. (Unanimously pass)

If the internationalization has made an organizational plan, and if the test team handles all strings (we think this can be automated), if the test team has made a manual test that facilitates localization function (we don't think this is effective Automated test), then a small set of automated test cases is to check whether localization is legal is needed and reasonable. We rely on local users to really use / manually test. (7 consent, 1 denial)

If the baseline and enhancement test are strong enough, the test script relaxed language is rarely done, except for a group of small tests, you need to select a script. (5 consent, 0 negation)

If translation is implemented, there is no early translatable design, then we need to retest everything. In this case, the baseline code may be a meaningful value for the automatic script.

We do not agree to the situation below: Repeat all the defects in the localized version, extension in baseline versions, and establish the same baseline to extend automation testing for each language. (7 disagree, 1 consent)

Bug classification may be a non-reliable reason to cause planning of the baseline recession test by baseline recession. (9 consent, 1 opposition)

We have not voted at these views because I will join Brian March recommends the discussion of the automatic localization test to join the test plan, and we want some easy distinguishing test classification. Low below is some examples:

Language is independent of automation test, such as (many of the cases not all cases) Print personnel configuration test, other configuration / compatibility tests, compatibility tests for changing paper size. Special language automatic testing if they are worth it. If you expect new products from France, the UK, Spain, then you might find value in some of the automated test cases of France, Germany, and Spain.

More use by manual localization test is the best way to test language details. The internationalization test is performed by automation processing. These test inspection software translation and localization. For example, you may provide a virtual very short string to translate. Different countries you might provide different connection text.

Cheap localization test. Marick's expectations are small. However, some tests do not care about string usage and screen display. For example, relying on constantly performing some functions to discover the pressure test of memory leaks, but the displayed graphics and text may have no relationship. The localization test in these tests, the bottom level is useful, which will be simple. The bottom line is even if you have a powerful automation test kit with English products, and you will not speed up the speed when you test the translation.

This article is completed, because the English level is limited, there is a lot of omissions, please forgive. I hope this article will give you some revelations. Thank you.

转载请注明原文地址:https://www.9cbs.com/read-23753.html

New Post(0)