English 中文(简体)
How to prove my stakeholder and manager my software works? [closed]
原标题:

What do software engineers encounter after another stressfull release? Well, the first thing we encounter in our group are the bugs we have released out in the open. The biggest problem that we as software engineers encounter after a stressfull release is spaghetti-code, also called the big ball of mud.

The time and money to chase perfection are seldom available, nor should they be. To survive, we must do what it takes to get our software working and out the door on time. Indeed, if a team completes a project with time to spare, today’s managers are likely to take that as a sign to provide less time and money or fewer people the next time around.

You need to deliver quality software on time, and under budget

Cost: Architecture is a long-term investment. It is easy for the people who are paying the bills to dismiss it, unless there is some tangible immediate benefit, such a tax write-off, or unless surplus money and time happens to be available. Such is seldom the case. More often, the customer needs something working by tomorrow. Often, the people who control and manage the development process simply do not regard architecture as a pressing concern. If programmers know that workmanship is invisible, and managers don t want to pay for it anyway, a vicious circle is born.

But if this was really the case than each longterm software project would eventually always lead to a big ball of mud.

We know that does not, always, happen. How come? Because the statement that managers do not regard architecture as a pressing concern is false. At least nowadays. Managers in the IT field very well know that maintainability is key to the business.

business becomes dependent upon the data driving it. Businesses have become critically dependent on their software and computing infrastructures. There are numerous mission critical systems that must be on-the-air twenty-four hours a day/seven days per week. If these systems go down, inventories can not be checked, employees can not be paid, aircraft cannot be routed, and so on. [..]

Therefore it is at the heart of the business to seek ways to keep systems far away from the big ball of mud. That the system is still maintainable. That the system actually works and that you, as programmer can prove it does. Does your manager ask you if you have finished your coding today, does she ask you if the release that has fixes A, B and C can be done today or does she ask if the software that will be released actually works? And have you proved it works? With what?

Now for my question:

What ways do we have to prove our managers and/or stakeholders that our software works? Are those green lights of our software-unit tests good enough? If yes, won t that only prove our big-ball of mud is still doing what we expect it to do? That the software is maintainable? How can you prove your design is right?

[Added later]

Chris Pebble his answer below is putting my team on the right track. The Quality Assurance is definitely the thing we are looking for. Thanks Chris. Having a QA policy agreed with the stakeholders is than the logical result of what my team is looking for.

The follow-up question is what should all be in that QA policy?

  • Having that buildserver running visible for my stakeholders
  • Having the buildserver not only just build but adding tests that were part of the QA policy
  • Having an agreement from my stakeholders on our development process (where developers review each others code is part of)
  • more..

Some more information: The team I m leading is building webservices that are consumed by other software teams. That is why a breaking webservice is immediately costing money. When the developers of the presentationlayer team, or the actual testers can t move forward we are in immediate stress and have to fix bugs ASAP, which in turn lead to quick hacks..

[ added later ]

Thanks for all the answers. It is indeed about trust . We cannot do a release if the software is not trusted by the stakeholders, who are actively testing our software themselves using the website that is consuming our webservice. When issues arise, the first question of our testers is: Is it a servicelayer problem or a presentationlayer problem? Which directs me to have a QA policy that ensures that our software is ok for the tests they are doing.

So, the only way I can (now) envision enabling trust with testers is to: - Talk with the current test-team, go over the tests that they are able to manually execute (from their test-script and scenario s) and make sure that our team has those tests as unit-tests already checked against our webservice. That would be a good starting point for a sign-off before we do a release that the presentationlayerteam has to integrate. It will take some effort to clarify that creating automatic tests for all those scenario s will take some time. But it will definately be usefull to ensure what we build is actually working.

最佳回答

You cant prove it beyond the scope of the tests, and unless there is a bulletproof specification (which there never is) then the tests never prove anything beyond the obvious.

What you can do as a team is approach your software design in a responsible manner and not give in to the temptations of writing bad code to please managers, demanding the necessary resources and time constraints, and treating the whole process as much as a craft as a job. The finest renaissance sculptors knew no-one would see the backs statues to be placed in the corners of cathedrals but still took the effort to make sure they weren t selling themselves short.

As a team the only way to prove your software is reliable is to build up a track record: do things correctly from the start, fix bugs before implementing new features, never give in to the quick hack fix, and make sure everyone shares the same enthusiasm and respect for the code.

问题回答

I m part of a team working on a large project for a governmental client. The first module of phase 1 was a huge disaster, the team was un-managed, there were nt a QA team and the developers were not motivated to work better. Instead, the manager kept shouting and deducting wages for people who didnt work overtime!!

The client, huh, dont ask about that, they were really pissed off, but they stayed with our company because they know that no one understand the business as we do.

So what was the solution:

  • First thing separated the management from programmers, and put a friendly team leader.
  • Second, get a qualified QA team. In the first few weeks, bugs were in 100s.
  • Third, put 2-3 developers as support team, there responsibility is not to do any new task, just fix bugs, work directly with the QA.
  • Fourth, motivate the guys, sometimes its not about the money or extra vacations, sometimes a good word will be perfect. Small example, after working 3 days in row for almost 15 hours a day, the team leader made a note to the manager. Two days later i received a letter from the CEO thanking me on my efforts and giving me 2 vacation days.

We will soon deliver the 4th module of the system, and as one of the support team i could say its at least 95% bug free. Which is a huge jump from our first modules.

Today we have a powerful development team, qualified QA and expert bug fixers.

Sorry for the long story, but thats how our team (during 4 months) proved to the manager and client that we are reliable, and just need the right environment.

In all but trivial cases, you cannot prove that your software is correct.

That s the role of User Acceptance Testing: to show that that an acceptable level of usefulness has been reached.

I think this is putting the cart before the horse. It s tantamount to a foot soldier trying to explain to the general what battle maneuvers are and why protecting your flanks is important. If management can t tell the difference between quality code and a big ball of mud, you re always going to end up delivering a big ball of mud.

Unfortunately it is completely impossible to "prove" that your software is working bug-free (the windows xp commercials always annoyed me by announcing "the most secure version of windows ever", that s impossible to prove at release). It s up to management to setup and enforce a QA process and establish metrics as to what a deliverable product actually looks like and what level of bugs or unexpected behaviors is acceptable in the final release.

That said, if you re a small team and set your own QA policies with little input from management I think it would be advantageous to write out a basic QA process and have management sign off on it. For our web apps we currently support 4 browsers -- and management knows this -- so when the application breaks in some obscure handheld browser everyone clearly understands that this is not something we designed the application to support. It also provides good leverage for hiring additional development or test resources when management decides it wants to start testing for x.

As Billy Joel once say, "It s always been a matter of trust."

You need to understand that software development is "black magic" to all except those who write software. It s not obvious (actually, quite counter intuitive) to to the rest of your company that many of your initiatives lead to increasing quality and reducing risk of running over time and/or budget.

The key is to build a trusting, respectful relationship between development and other parts of the business. How do you build this trust? Well that s one of those touchy-feely people problems... you will need to experiment a little. I use the following tools often:

  1. Process visibility - make sure everyone knows what you are doing and how things are progressing. Also, make sure everyone can see the impact of and changes that occur during development.
  2. Point out the little victories - to build up trust, point out when things went exactly as you planned. Try to find situations where you had to make a judgement call and use the term "mitigated the risk" with your management.
  3. Do not say, "I told you so." - let s say that you told management that you needed 2 months to undertake some task and they say, "well you only have three weeks." The outcome is not likely to be good (assuming your estimate is accurate). Make management aware of the problem and document all the things you had to do to try to meet the deadline. When quality is shown to be poor, you can work through the issues you faced (and recorded) in a professional manner rather than pointing the finger and saying, "I told you so."

If you have a good relationship with you manager, you can suggest that they read some books specific to software development so they can understand industry best practices.

Also, I would point out to your boss that not allowing you to act as a professional software developers is hurting your career. You really want to work somewhere that lets you grow professionally rather than somewhere that turns you into a hacker.





相关问题
run unit tests and coverage in certain python structure

I have some funny noob problem. I try to run unit tests from commandline: H:PROpyEstimator>python src estpython est_power_estimator.py Traceback (most recent call last): File "src est...

How to unit-test an enterprise symfony project?

I´m working on a huge project at my work. We have about 200 database tables, accordingly a huge number of Models, Actions and so on. How should I begin to write tests for this? My biggest problem ...

Code Coverage Tools & Visual Studio 2008 Pro

Just wondering what people are using for code coverage tools when using MS Visual Studio 2008 Pro. We are using the built-in MS test project and unit testing tool (the one that come pre-installed ...

Unit testing. File structure

I have a C++ legacy codebase with 10-15 applications, all sharing several components. While setting up unittests for both shared components and for applications themselves, I was wondering if there ...

Unit Testing .NET 3.5 projects using MStest in VS2010

There s a bug/feature in Visual Studio 2010 where you can t create a unit test project with the 2.0 CLR. https://connect.microsoft.com/VisualStudio/feedback/ViewFeedback.aspx?FeedbackID=483891&wa=...

Unit Test for Exceptions Message

Is there a simple (Attribute-driven) way to have the following test fail on the message of the exception. [TestMethod()] [ExpectedException(typeof(ArgumentException))] public void ExceptionTestTest() ...

热门标签