meta user
network profile
| bio | website | testingmentor.com |
|---|---|---|
| location | Seattle, WA | |
| age | ||
| visits | member for | 2 years |
| seen | May 12 at 15:39 | |
| stats | profile views | 35 |
Principal Test Lead in Microsoft's Windows Phone team.
|
Jun 3 |
comment |
Has automated UI regression testing helped you in finding critical defects in system? You're correct, that perhaps was a faulty assumption on my part. From your anecdotal statements "tests failing intermittently" and "without clear reason" I jumped to potential conclusions for these types of issues might include sync/timing/race conditions in your script, and lack of diagnosability (e.g. expect "foo", actual "http 403", and stopwatch function). My comment was not intended to disrespect your automation project, but was simply to counter your statement that automation can never find new bugs. |
|
Jun 3 |
awarded | Supporter |
|
Jun 3 |
comment |
Building “slow to break” regression tests Despite its drawbacks, there are also benefits of UI automation. However, to argue that automating UI testing is "a degradation of the testing process" blindly ignores the contexts in which it can be beneficial. As a side note: sapience is "having or showing great wisdom or sound judgement." Useful UI automated tests that provide value requires a sapient tester to design and develop, as does behavioral and explorative types of testing. Of course, not all automated tests are developed by sapient testers, but humans "trying and seeing" also doesn't necessarily reflect sapience. |
|
Jun 3 |
comment |
How to 'Mock' SVN @Dale, the cost factor is in ref. to the cost of setting up SVN in a clean room env. for testing purposes, versus the cost of coding and maintaining mock objects. I understand value of mocks esp. in unit testing, and also the risks (e.g. mocks model functionality, and may provide false negatives) Also, at some point the mocks have to be replaced with the real thing. IHMO, in this case it is 6 of one thing and half dozen of another since we don't really know complete story. :-) |
|
Jun 2 |
answered | Are state transition diagrams a good software testing technique |
|
Jun 2 |
awarded | Critic |
|
Jun 2 |
comment |
How to 'Mock' SVN When I first read this question, I asked myself why would anyone want to 'mock' SVN? The purpose of a mock is to simulate an object when the 'real' object is not available or impractical to use. IMHO, Writing a mock of SVN doesn't seem to be a cost effective solution since you can get SVN for free. As with Dale, I am confused over who is throwing and who is catching (or rethrowing) 'the exception.' |
|
Jun 1 |
answered | MTBF for a software product |
|
May 28 |
awarded | Editor |
|
May 28 |
revised |
What types of testing are LEAST suited to using pairwise testing (and similar test design approaches)? Clarification |
|
May 27 |
answered | Is a distinction between “testing” and “checking” useful? |
|
May 27 |
suggested | suggested edit on What types of testing are LEAST suited to using pairwise testing (and similar test design approaches)? |
|
May 27 |
awarded | Teacher |
|
May 27 |
answered | What types of testing are LEAST suited to using pairwise testing (and similar test design approaches)? |
|
May 6 |
comment |
Has automated UI regression testing helped you in finding critical defects in system? Your example is not a failure of test automation, it is an example of a poorly designed autoamted test with poorly designed oracles. I can give multiple counter examples of where randomized inputs using random test data generation and data-driven combinatorial testing found many bugs in GUI automated tests. |
|
May 6 |
comment |
Has automated UI regression testing helped you in finding critical defects in system? If your automated tests are brain-dead scriptlets with a bunch of hard code actions/data, then I agree with your statement. However, to claim that GUI test autoamtion will "never find NEW bugs" is simply incorrect. Well-designed automated tests can and do find new bugs. |
