Testing whole programs. best practices -
I am currently developing a library and a set of programs using this library in Python. Unit testing indicates that I import each module from the library and test it within the classroom and routines. no problem with that. I have a separate test directory in which all of my testing and library modules are imported, which I run when developing.
However, when it comes to testing programs, things change. To test, the program must run as a whole program to believe in establishing the library (which may actually be the case, though incorrect, if I have installed its previous version on my machine, then the next problem Add). At the moment, my programs are run by a streetview with a pithpath definition, which I do before (IOW, I do not install) handling, but I do not think I am correcting it. I generally feel as if a program is fully deployed, it should be tested for efficiency, but this will mean that I have to set it to perform a functional test every time.
Your experience and suggestions Functional testing of the entire program? Do you do before or after the deployment, and how?
Thank you
Note that I do not include Python tags on purpose although my problem is python specific, and I would like the answers related to Python, I think the contribution Experts from other languages can also be brought from. Reported in a comment as
Edit : In fact, my program, when installed, has to import those modules whose path Can only be deployed (I can download and install dependencies on the fly, they are not installed on my machine). I can not manipulate sys.path from the test, because this would mean that I modify a program (my executable) with a program (my executable) (Trituit, which is running and a system call) Generates).
In other words, the only way to test the program without deployment is to set up a program along with Pythonpath in Dir and Library in which the program script (which, as I said, Download, compile and install "Everything" in a temporary directory.)
On deployment, Depp and executable are packaged in an "OSX bundle" structure, which Uri way executable
Edit :
It has been added a 150 reward to see if I can get some more feedback.
Edit :
I appreciated all the answers and voted them all. The election is a tough call for me, but I have been reminded about the V-model approach by LudomCo which I had read long ago. Thanks for everyone for a very good answer.
In our company, we are using a very commonly used V - forms of development process In the model where the unit is carried out within the implementation phase, the integration test is done against the architecture / design phase, and there is a system test against the stage of the requirements.
So in your case, what I think, you want to test your application thoroughly at a functional level. Therefore, it has to be done against the requirements.
In this way, you need the necessary documentation, whether full text scenario or (better) case diagram of a UML use (ideally large scale) application (usually from the first stage of receiving A) Then you have to write the test cases covering the case of each use and you have to pass these test cases. This can be done through manual or automatic testing with famous (or very expensive) devices.
When , we usually test these systems after deployment (using the test team is the installer provided by the developer team), as we also install the installer , Where the integration based on the case occurs before or after the trial or deployment.
In your case, if the installer is without error, and you are 100% sure to test using your PYTHONPATH variable before being deployed, once deployed it will never bring a bug, You can then choose to test before deployment. This is pure risk management and this is your call, because you are the one who knows its best and the opposition for its applications. (Personally, I do not know why insects are not there, they are everywhere :-))
Hope that helps and I'm not an off-topic
Comments
Post a Comment