Motorhome App (1 Viewer)

[GJH]

I am bemused by this little tiff. The proof of the pudding will be in the eating. If it works and does what Grommet and others want then how he achieved that is entirely up to him isn't it? Or have I missed something?

I always judge by results, but then I come from the construction industry. And so the only thing I know about the IT industry is that it is the only other one I know that makes the construction industry look good at project management.

It's like any other engineering project. You don't start to construct a building until you have designed it. You don't start the design until you have specified what the function of the building is to be and where its various components are required to be.

Interestingly enough, classic information engineering techniques are based on architectural practice. In a building, for instance, it doesn't really matter if you decide that a bedroom is to be used as a playroom or a study. However, if the bathroom and kitchen have been put in the wrong places then rerouting all the plumbing &c is a much bigger job - so it pays to do the analysis and design thoroughly before starting to build.

Karl asked earlier "How can you design a test plan when you don't even know how the software works internally." The point is that you don't need to know how the software works internally. What you do need to know, however, is what results should be produced by the software for any given inputs because only by doing so do you know whether it is working. It's the same sort of thing as testing a structure like a bridge. It doesn't really matter how it is built but it does matter that it is capable of taking the weights it is supposed to take.

 

[DBK]

Great stuff!

What, the arguments?

 

[DBK]

It's like any other engineering project. You don't start to construct a building until you have designed it. You don't start the design until you have specified what the function of the building is to be and where its various components are required to be.

Interestingly enough, classic information engineering techniques are based on architectural practice. In a building, for instance, it doesn't really matter if you decide that a bedroom is to be used as a playroom or a study. However, if the bathroom and kitchen have been put in the wrong places then rerouting all the plumbing &c is a much bigger job - so it pays to do the analysis and design thoroughly before starting to build.

Karl asked earlier "How can you design a test plan when you don't even know how the software works internally." The point is that you don't need to know how the software works internally. What you do need to know, however, is what results should be produced by the software for any given inputs because only by doing so do you know whether it is working. It's the same sort of thing as testing a structure like a bridge. It doesn't really matter how it is built but it does matter that it is capable of taking the weights it is supposed to take.

Unless I've mistaken Gromett's intention he is making a sort of intelligent note book, with various forms and things to be filled in and various reminders. It isn't a classic data processing programming so I am unsure what test regime needs to be designed beforehand?

Of course I speak as someone whose only qualification in the subject was a four week course in something, if my memory isn't totally shot was called SSADM. I've no idea why I did it but it was fun up until the point when it got to TNF or Third Normal Form at which point my brain melted and dribbled out of my ears.

I hasten to add this was a long time ago at the School of Signals at Blandford and during the course the great hurricane struck, the one that made Sevenoaks into Threeoaks. We watched caravans in storage being blown by the wind like tumbleweeds across the Regimental parade ground.

^{_{Subscribers  do not see these advertisements}}

 

[Gromett]

It's like any other engineering project. You don't start to construct a building until you have designed it. You don't start the design until you have specified what the function of the building is to be and where its various components are required to be.

Interestingly enough, classic information engineering techniques are based on architectural practice. In a building, for instance, it doesn't really matter if you decide that a bedroom is to be used as a playroom or a study. However, if the bathroom and kitchen have been put in the wrong places then rerouting all the plumbing &c is a much bigger job - so it pays to do the analysis and design thoroughly before starting to build.

Karl asked earlier "How can you design a test plan when you don't even know how the software works internally." The point is that you don't need to know how the software works internally. What you do need to know, however, is what results should be produced by the software for any given inputs because only by doing so do you know whether it is working. It's the same sort of thing as testing a structure like a bridge. It doesn't really matter how it is built but it does matter that it is capable of taking the weights it is supposed to take.

Ahhh, That bit makes sense. Yes I have done all that. I have even gone through various tests of UI layouts to ensure that I don't get half way through and realise it won't work the way I was doing it.

What I haven't done is specified which features belong to which module. I haven't done the database table layouts yet or normalisation, I haven't specified which classes are needed and what functionality is required by each. I haven't done the inheritance hierarchy layouts simply because I don't know what classes are provided by java or Android. I haven't decided which functionality I need to write because some will be provided by Android. etc etc etc.

I have however created a great long list of features I want to provide. I have then split these into clearly defined modules. I have then designed a UI interface layout that will be easy for the user and will allow me to separate the functionality of the app out into separate modules. Within each module before I start writing I sit down with a note pad and decide what I want to achieve at each stage, I do rough drawing of how it will look and imagine using it. I have been using software long enough now that I can imagine the process pretty well and make notes of any gotchas.
I then write down any data structures I think I will need including all elements and their type/size. For each data structure I will rough out a list of functions needed for the elements of the structure such as getters, setters and validators. I then also look to see if there are any design patters I can bring to bear without over complicating things.

Once that is done I will have a few pages of notes and will then crack on with banging in some code.

It is not a formal process and it is not structured documentation. It is just a programmer planning ahead so he doesn't trip himself up.

I thought we were at crossed purposes here. In architectural terms, I have decided what the building is for, The dimension of each room, where they are in the building, where the stairs and passageways are. What I haven't done is designed a full electrical, plumbing layout and specified the wire sizes, pipe sizes, what size RSJ is need for each structural element etc etc. Currently I don't know enough about the plumbing and wiring of Android to be able to specify those elements.
For instance until yesterday or was it the day before I didn't know about menu inflaters. So Prior to that I would have probably have written a custom menu class that disabled elements based on the Activity I was creating. Now I know I just have to create a custom inflater for each Activity. This has changed the structure of the program and removed a couple of classes and at least 1 file from the project.

As I said, I thought we were talking at crossed purposes. I don't just sit down and start banging in the code without any prior thought or planning. Only idiots do this.

 

[Gromett]

Unless I've mistaken Gromett's intention he is making a sort of intelligent note book, with various forms and things to be filled in and various reminders. It isn't a classic data processing programming so I am unsure what test regime needs to be designed beforehand?

There is a test regime needed. Basically where ever I am passing or receiving data to/from an external module or function I prefer to validate that data. It is a security issue as much as a bug prevention issue. That is the main function of my testing processes. To be honest if the program crashes it is not critical. It is not in a hospital nor is there any lives depending on it. However if my software was the source of a security issue I would be seriously ashamed of myself as a developer.

 

[peterc10]

It's like any other engineering project. You don't start to construct a building until you have designed it. You don't start the design until you have specified what the function of the building is to be and where its various components are required to be.

Interestingly enough, classic information engineering techniques are based on architectural practice. In a building, for instance, it doesn't really matter if you decide that a bedroom is to be used as a playroom or a study. However, if the bathroom and kitchen have been put in the wrong places then rerouting all the plumbing &c is a much bigger job - so it pays to do the analysis and design thoroughly before starting to build.

Karl asked earlier "How can you design a test plan when you don't even know how the software works internally." The point is that you don't need to know how the software works internally. What you do need to know, however, is what results should be produced by the software for any given inputs because only by doing so do you know whether it is working. It's the same sort of thing as testing a structure like a bridge. It doesn't really matter how it is built but it does matter that it is capable of taking the weights it is supposed to take.

If you really do base your processes on the way that architects (mis)manage building works then I can now understand why so many IT projects fail.

^{_{Subscribers  do not see these advertisements}}

 

[GJH]

If you really do base your processes on the way that architects (mis)manage building works then I can now understand why so many IT projects fail.

The architectural bit comes into the requirements and design. Project management is a separate discipline, no matter what the field.

 

[berni109]

KISS always works best.
A small hobby project thats the result of personal endeavor while learning new skills will probably (almost certainly) come out better than all of the formal processes could hope to achieve.
A team of development professionals working on medium to large projects will never succeed in delivering a sustainable product without formal processes.
Horses for courses!

 

[GJH]

KISS always works best.

The golden word - and one of my first when training programmers (i.e. making them unlearn all the bad habits picked up from lecturers) in the dim & distant past

^{_{Subscribers  do not see these advertisements}}

 

[scotjimland]

@Gromett

what's the latest news on the project Karl ..

dead in the water or still a 'work in progress' ?

 

[Gromett]

@Gromett

what's the latest news on the project Karl ..

dead in the water or still a 'work in progress' ?

Still a work in progress. Summer got in the way and I ran out of funds a month or so ago so I am having to get some work in. Come November the project will start moving again I think. I should be able to get a load done over Christmas/New year however as I won't be able to get any work in I don't think. Should have 2-3 weeks of nothing else to do... However hoping to have a module or two done by then,

I have the basic UI done and working already.