Toughest Testing Challenge 1.2 : Patch 1.0

There was a major bug in Toughest Testing challenge 1.2 .

Bug Details:
Management realized later that a simple module (dialogue box) like this can’t be tested forever, also they don’t have any idea about how far or how long to test this module.

So, the management decides to give the project to  two different vendors.
First one was asked to generate  test ideas that can be executed within 15 minutes.[Time is non negotiable here]
A second vendor was asked to estimate the time required to test this module and generate test ideas for the time estimated.

What would be your test strategies if you are Vendor one?
What would be your test strategies if you are Vendor two?

And more over you should convince the client with your test strategy and estimates.

A Hint : Test Framing.

Happy Testing!

Dhanasekar S.

Norie’s Neat Nostrum :
“There’s no such thing as quick and dirty; if you want a quick job, make it a neat job” – Jerry Weinberg

Think different but keep it simple :)

There is content below, I can read can’t you? oh!  Think different to see the content, but keep it simple 🙂

This is the most common problem we testers encounter quite often, it works in developer machine but not in test machine.

Once an interesting incident happened in one of the web application I was testing, application was working fine in developer’s environment but not in test environment.

The most likely reasons could be either  the test servers do not have the latest code or  there is conflict in environment. Quick check revealed that both the environments had the same code.

Now the focus was on to investigate the configuration of the servers.

As a tester how will  you investigate such situations?

Few tips to nail down such problems. Please share your tips as well.

1. Test in another test machine\server

2. Test in another dev machine\server

This will help to narrow down if there the problem is at client side or server side.

3. Check if  that machine  is also hosting the web server.

Time out issues and session management related issues may not be reproducible, if the machine you are testing in is a server machine.

4. Check for additional softwares installed.

Generally developer box will have lot of additional softwares installed, like debug tools, development related softwares that may prevent from the bug to be reproduced. Even the web browser in dev box will be loaded with tons of additional tools and utilities. Such additional installation may also prevent from some bugs being reproduced.

It is not enough to check if required softwares were  installed, it is equally important to make sure no additional unwanted softwares were installed.

Coming back to the issue, following those above mentioned approaches by pairing with developer helped to nail down the problem in few hours. The issue was, a bulk data extraction component used (it was ETL project) will work only in localhost, that does not have the capability to connect with external host. In the developer machine and dev server both web server and DB server was hosted  in the same machine, so that worked perfecly fine.But test environment had DB and Web server running in different machine, so this caused the problem. An important lesson learned from the incident is it is always good to have the test environment mimic the production. Keeping the test environment similar to production helped in finding the problem very earlier in the cycle.

Another good practice to follow, at any given time at least one test environment should be  in sync with production server, so any production issues reported can be easily replicated and analysed in a test environment. Having only one test environment means most time that will be updated with enhancements going on, that may also prevent from reproducing production issues because of code changes applied in the test environment.

How many of you have set up the test environments? No, I didn’t mean double clicking an installer provided by the developer. As a tester I learned a lot, by setting up test environment and maintaining it, setting up test environments will help  to understand your application better, it will help to get better test ideas, analyzing of bugs can be done  effectively .

Understanding test environment improves testing.

——————————————————————End of Post—————————————————————————–

Test[ing this] Case

The first thing flashes in most software testers mind immediately on hearing software testing is Test Case. Even most of us explain our work to others and even to fellow testers something like this “I understand requirements, write test cases and test the application, log the bugs and mark the test case pass or fail based on the expected results”. I too started my testing career by explaining what a test case is and SDLC model that was (never) followed to an interviewer. But soon, I realized that it is practically not possible (or I am not intelligent enough) to write most test ideas (cases) even with detailed requirements. Most test ideas don’t popup while analyzing the requirements, but while performing the real testing on the product. The test cases written with requirement documents are only based initial abstract understandings made on theory, but the real practical ideas come while testing. But even recently many people advice to write detailed test cases, that even a layperson can understand. Are they recruiting a layperson or a smart tester? Why still the main focus is with writing detailed test cases even in agile world? Looking back the water fall model, what I can think of reason for writing detailed test cases is, there was  a phase in SDLC were developers are busy writing and developing a product for months together, before delivering into testing team. So, probably some smart manager got this idea of test case writing to keep the testers engaged during that phase. Few rituals still followed without analyzing, if those suit the context. 1.  Write detailed test cases for the projects where there is good clarity on requirements. We got clear requirements so write detailed test cases. 2.  Write detailed test cases where there are no clear requirements . The requirements are not clear let use have detailed test case document. 3.  Even if the project has very tight deadline, a certain percentage of time is allocated for test case writing instead of focusing on real testing . Because Test case documents are (mis)believed to be the only Savior of testers. 4.  I’ve even seen reverse engineering of coming up with test cases from a well functioning application, with the help of help files 🙂 5.  There is something called traceability matrix, which traces all the detailed test cases back to your requirement, I think this is designed especially to say the whole world, see we’ve spent so much of man power to duplicate the requirements and here is the proof for that 6. And test cases are again considered as Savior in this job hopping era, but reality is forcing to write such detailed test cases and asking the testers to blindly execute just the documented test cases too are the also one of the reasons for job hopping. Why a similar kind of test cases is demanded everywhere for every context?Why do test case document looks similar where ever I go? 😦 . The only standard that is followed more or less correctly in testing world is test case template. The problem with such detailed test cases writing and following them is, soon the application will develop resistance against bug, because testers were focused only on what is written in the test case nothing less nothing more. Any tests executed blindly by a human just against expected result is as good as (as bad as) automated test checks. Such practices never help to spot the Black Swan* [The disproportionate role of high-impact, hard to predict, and rare events that are beyond the realm of normal expectations] you have to explore more with real application to spot the Black Swan. Testing according to James Bach Testing is the infinite process of comparing the invisible to the ambiguous so as to avoid the unthinkable happening to the anonymous. So, let me try to form a different Black Swam theory, as a tester our job is to predict and spot the Black Swan before the customers or client spot it. 🙂 Even writing test cases need exploration, but it becomes a concern when the explorations stop (or forced to stop) once test case are written. So, constantly come up with new ideas to add more value. Instead of wasting time in writing detailed test cases cheat sheets like this  Test Heuristics Cheat Sheet would cover most of the basic test cases, why to write again? If you feel You Are Not Done Yet, then check this You Are Not Done Yet. And choose SBTM or any relevant format that suits the context. Any available basic requirement document plus the above cheat sheet should be more than enough to do a good enough testing.

Testers aren’t  for documenting, they are for providing valuable information.

P.S: Can you recollect  the color of each letter of Google logo in its home page? (Of course with no doodles), the page most of  us visit every second hour. If you can’t, then you are affected with major Black Swan blindness.

Happy testing your Case!

Dhanasekar S.

* I’ve just started reading The Black Swan book by Nassim Nicholas Taleb. Yet to read it completely.

Configure, Operate, Observe and Analyze

In one of my previous post I wrote about the below Windows  OS (XP with SP2) error message. This is one of very interesting behaviors that took me few hours to crack down.

As a novice Window’s(OS) user.

Access is denied.

Ah!  I am the only user of my machine, and I am the one who created this folder, no one else has any rights to restrict me from accessing my own folder.

Make sure the disk is not full or write-protected.

What the hell! Don’t say all this nonsense. I didn’t protect it from writing. I can write whatever I want in the disk. How do I know the disk is full? How do I write-unprotect? Or an unwrite-protect 😉 ?

That the file is not currently in use

No, I am not renaming any file. I am renaming a folder, what is your problem?

User annoyed and calls customer care.

Now, as an advanced computer user.

Access is denied.

Looks like the permission settings for the folder is changed, may be my wife? Son? Or Pet? Possible. Let me see the permission setting for the folder. Oh! This user has full access. Then, what is the problem?

Make sure the disk is not full or write-protected.

Nope, my drive has enough space, not write protected.

That the file is not currently in use

Files? I am renaming a folder. Let me see if the folder is kept open. Yes, so let me close and try. Still I am getting the same error.

Annoyed user might solve the problem by logging off or closing all files and folder or restarting the machine, which forces him to close all the programs kept open.

As a testers, start analyzing (Step 4 of testing Analyze, Step 1: Configure, Step 2: Operate, Step 3: Observe) the error message and follow the error message to see, to find the problem area.

Access is denied.

Does the application provide reasons to the Error? Yes it is it says

Make sure the disk is not full or write-protected.

The basic disc properties say it is either full or write-protected.

That the file is not currently in use

No, I am renaming a folder, but still let me check if the folder is  kept open? Yes.

So, let me close the folder and rename it. Even after that I am getting the same message.

Are there any files inside the folder?

Yes, there are few.

Are they open?

Yes, a word document inside that folder is kept open.

Is that causing the problem? Let me close it and try.

At last I can rename it, so “When a file inside a folder to be renamed is kept open, the user is not show with an appropriate error message” the problem summary?

I have just solved the problem but haven’t got the exact steps to reproduce.

The problem was identified without knowing the proper base step up or configuration. So let me go to base state, also let me note down or take a snap shot of the current state where the problem was first spotted, including other files\folders those kept open, processes running to name a few.

So, what is the base state? I may even restart my machine or log off or close all opened files and folders.

Step 1: Configure

Let me log off and log in. Note down the entire process running (and use some tools like Systracer that takes a snap shot of your machine)

Step 2: Operate

1.    Open the Disk Drive (Say D :\)

2.    Open the folder (Say D:\Test)

3.    Open a file inside the folder (Say D:\Test\test.doc)

4.    Use back button to get back to disk drive

5.    Try Rename the folder

Step 3: Observe

Ah! This time I am getting a proper message.

So, what could be the problem?

Let me try changing the sequence, let me do this

1.    Open the Disk Drive (Say D :\)

2.    Open the folder.(Say D:\Test)

3.    Open a file inside the folder. (Say D:\Test\test.doc)

4.    Close the folder

5.    Reopen the folder

Now the sequence of opened windows changed, now file first then the folder.

Nope! I got proper message. Tried with different types of file kept open like excel, txt files, folder within folders and also be keeping few more similar file types open in same\different folders.

With no success decide to get back to the dirty environment and try again, I opened few other windows ,files, browsers and application those were kept open when I first found this issue, and tried all the above steps, this time I noticed, I was  able to reproduce the error once in a while but not consistently. But after more than an hour I was not able to find the exact steps or pattern to reproduce it.

Then decided to quit and try later (Plunge and quit), but that one hour of exploration gave me confidence that it can be reproduced.

Few days later with lots of windows and browsers kept open I tried again to reproduce this. Initial few attempts were not successful.

When I was about to rename the folder, I noticed one of my friends pinging me in IM, so I went to the browser and replied to him, then came back to folder window to rename the folder.

Whoopee! I got error message I am looking for this time. It looks if I navigate between different windows I might be able to reproduce, but even trying that didn’t reproduce the error consistently. What next?? Trying out few other ways and finally was able to narrow down the steps to reproduce it consistently.

And, Here are the steps to reproduce.

1.    Open the Disk Drive (Say D :\)

2.    Open the folder (Say D:\Test)

3.    Open a file inside the folder  (Say D:\Test\test.doc)

4.    Use back button to get back to disk drive

5. Wait (approximately) 10 seconds

6.    Try Rename the folder

7.    Bingo!

Note to Readers:

  • Heuristics applied to solve are of James Bach’s Learning Heuristics SACKED SCOWS

Scouting Obsessively – discover the sources and tool you need
Authentic Problems – engage the mind
Cognitive Savvy – work with the rhythms of the mind
Knowledge attracts Knowledge – the more I know the easier I learn
Experimentation – make learning vivid and direct
Disposable Time – lets me try new things (I tried this in my disposable time)
Stories – are how I make sense of things
Contrasting Ideas – lead to better ideas (Skepticism, Critical thinking, Lateral thinking, Systems thinking)
Other Minds – exercise my thinking and applaud my exploits
Words and Pictures – make a home for my thoughts
Systems Thinking – helps me tame complexity

Also, Procrastination , Plunge In and quit.

  • How to reproduce is solved. Now find why this happens?
  • Never let the user to get annoyed. As a tester, analyze every behavior of your application.
  • Read How to Investigate Intermittent Problems to learn the ability and the confidence to investigate an intermittent bug.
  • “There are no accidents” –Master Oogway, Kung Fu Panda(Movie)

FaceBook Logoff

I have this habit of logging into any shopping site like Flipkart with same ID, same time in two different browsers (Firefox and Chrome). With one of the sessions, I browse through the items, and other one to add to the shopping cart. This is very helpful to me, to do shopping faster.

I thought I would do the same with Facebook as well. I would use one of the sessions to read all the status updates and other session to chat or render the videos and watch, so I can browse through all the updates from friends in quick time. But I observed logging into second browser forces to log out from the previous browser session

Is this a problem?

I don’t think as a tester you can decide if this is a really a problem assuming either one of those scenarios as an industry standards ;-).Even if one of those scenarios is set as an industry standards it is not necessary to follow them blindly. So Face off 1 yourself as a user to analyze such scenarios.

We cannot think of everything before we start.

You don’t know what you want until you see it.

While shopping user (like me) has two sessions opened. I am done with my shopping .I do check out after adding all the items to the chart and make payment, check my mailing address,which  takes close to 5 to 10 minutes. And then I log out of this session but completely forgot about the other session opened in different browser, and if the user is using a shared computer which is very common in India. So, here it looks it is better to log off all the sessions in the machine if user log off from a session. But one way it is helpful to user if he can access more than one session at a time.

So, which is the correct behavior? It is always decided by what your users are up to. So face off like a user and analyze the observed behavior.

Even let say Flipkart decides to change this and allows only one user session similar to Facebook by maintaining the sessions in server instead of cookies. It should be communicated to user in a very positive manner by highlighting the security risk. Any change like this cannot go without educating the user. Users who are very used to such behavior may not accept those changes.

User satisfaction comes from handling change, not mitigating risk.

Wait. Did I say it is the user always? No(w), I am seeing some new terminology being used these days apart from user friendly, its advertiser friendly ;-). Sometimes you have to consider advertisers as well, so no harm in annoying user for some time.  There you should not think about user friendly close buttons.

Aware of your potential user\customer very well, to analyze what you observed is really a problem. None of the so called industry standards or testing techniques like orthogonal array, boundary value analyses or whatever it is, will help to capture or analyze such scenarios, and we can’t estimate for such analyses in advance. So

Apply your sapience and heuristics. Don’t follow rules.

“Hell, there are no rules here, we’re trying to accomplish something” ~ Thomas Edison

Note to Readers:

  • Here I have given just one possible scenario that I generally do as a user, I like you to bring\think in different perspective to analyses.
  • I had a vague idea about this log off session, was thinking on how to shape this in to an article. Finally this got this shape after  reading this wonderful post by Nathan Smith Estimation is bunk

1 –  Face off (movie) – A revolutionary medical technique allows an undercover agent to take the physical appearance of a major criminal and infiltrate his organization.

Simple application, Simple Bug(?) and a Simple Investigation

NOTE : If you ask how simple is it? it depends 😉

Twitter changed the way I communicate. It is simple, sweet, short and faster way to share information.

Recently I started using HootSuite, a social media tool to manage my social networking activities. This has a nice little option to view twitter profiles. On clicking a twitter handle opens a pop up that displays bio, picture options to reply, DM, follow, unfollow and few others. I clicked on my own twitter profile. To my surprise that also has follow, unfollow, DM options enabled. So first tried to Follow myself (a tester in full swing 😉 ) got a message “You can’t follow\unfollow yourself” hmm that looks good.

Then, I clicked on Direct Message(DM) to myself and was able to send a DM successfully to myself. I am not sure if this is the correct behavior or is a problem? If it is a problem, where is the problem?

I never tried sending DM to myself from twitter, so next tried to send DM to myself from twitter.

From the Direct Message options, tried sending a DM, it was not allowed.

Next tried with other option available to send DM, home page directly.

Oh! it is possible and the DM delivered to my twitter inbox. So there is inconsistency in handling DM to oneself in twitter. So, it is a problem.

So,as  a tester should look for as many details as possible like trying out all possible input and output options at API level, DB levels and UI levels, before logging a bug. That will help report the problem clear thus issue might get fixed faster and better.

Above all, first thing you should do on finding an issue is check if that is already logged. If yes, compare with your observation with those details, and provide information if you feel something is missed out.

Exploring more on this, Twitter help on DM says DM cannot be send to a user who is not following you and there is no mention of DM to yourself.  Since I can send a DM to myself but I am not following myself, so does this mean the claim made by twitter that “you cannot send a DM to a user who is not following you” is wrong?

What is your view?

Now your turn to investigate the not so user friendly message.

Below error message was from Windows XP with SP2, this happened while renaming a normal windows  folder. As a user what did you understand from this message? And what actions would you perform as a user to solve this problem?

Here is the investigation report of above Windows bug

Click here to see more such bug(?) posts.