Design and Architecture

Re: What's The Right Way To Do This?

22-Jun-12 10:56

I got to thinking about my design some more....

My application will be installed in a retail environment. My initial design calls for the software to work even if there's a loss of internet connection. So I thought of installing a local copy of SQL for my app to work with. Each night the app's update service would then transmit any data to the server that's not already been sent.

But then I got to thinking, it's possible for there to be multiple PC's in any given store. What if there's no network? I could offer to install a basic peer to peer network, and make the first PC the server, then all the other PC's could hit the DB there.

I went into this thinking thatI can't assume an internet connection. Do you think this is unreasonable in today's world? Should I build the app so that all data resides in the server and require them to have an internet connection? It would solve both the internet and network the issues.

Your thoughts?

Any thoughts on this?

If it's not broken, fix it until it is

Mycroft Holmes22-Jun-12 13:10

Mycroft Holmes

22-Jun-12 13:10

It depends on your market, small outlets I would think. However even small outlets often have multiple POS. A large, multi shop company will require the data to be consolidated but the POS operation is decidedly local.

I would build a local environment (WPF on the desktop) and make sure you cater for a replication scenario where the local server can replicate its data to HO. Personally I see absolutely no requirement for the internet in such an environment.

Never underestimate the power of human stupidity
RAH

Re: What's The Right Way To Do This?

Kevin Marois22-Jun-12 14:01

22-Jun-12 14:01

Ya, it's a tough design decision. I'll sleep on it over the weekend.

Thanks

If it's not broken, fix it until it is

color mixing riddle

Kobi_Z13-Jun-12 21:52

13-Jun-12 21:52

hi, my question is
if i have 3 basic colors (each made of rgb):
color1 : R:150, B:zero, G:255
color2 : R:255, B:150, G:zero
color3 : R:zero, B:255, G:150

each of them can be mixed using the formula :
new_color = floor(X*0.9)+floor(Y*0.1)
X and Y can be abasic color or a new color allready created by using the formula.

for example, if i want to mix color1 as main with color3 :
new_color(R,B,G) = (floor(0.9*150)+floor(0.1*0) , floor(0.9*0)+floor(0.1*255) , floor(0.9*255)+floor(0.1*150) ) = (135, 25, 244).

<b>I need to find a way to mix the colors in order to get a desired color, for example : R:187 B:135 G:201.</b>
so far i wrote a "brute force" program which go all over the combinations of basic colors (runing for 7 days now...)

hope there is a smarter way to solve the problem.
Thanks.

Bernhard Hiller13-Jun-12 22:18

Bernhard Hiller

13-Jun-12 22:18

That's called linear equations.
150a+255b+0c=187
0a+150b+255c=135
255a+0b+150c=201

Kobi_Z14-Jun-12 4:52

14-Jun-12 4:52

by solving this equation i get an answer :
X = 0.6077
Y = 0.3759
Z = 0.307

That give me nothing...

Im talking about mixing many basic colors in order to get one special, i need to find the mixing sequence.

Philippe Mori17-Jun-12 2:50

Philippe Mori

17-Jun-12 2:50

At that point, the problem is half-solved. Next step would be to find how to get there...

One way, would be to mix Y and Z (and any resulting color) until you get the proper ratio by using the best 2 candidate colors that consist of Y and Z or their mix.

Once you have done that, you repeat same process with that mix and X.

The ratio for Y and Z alone would be 0.55 and 0.45.

1) 0.9Y + 0.1Z --> Mix 1
2) 0.9 M1 + 0.1Z --> Mix 2 --> 0.81Y + 0.19Z
3) 0.9 M2 + 0.1Z --> Mix 3 --> 0.72Y + 0.28Z
...
6) --> 0.53Y + 0.47Z

At that point, you would take 0.9 of that and 0.1 of Mix 3... and you get 0.549 of Y.
If you need more precision, you can do more mix...

Philippe Mori

Richard MacCutchan14-Jun-12 5:19

14-Jun-12 5:19

Since you know the final RGB values that you need, what is the purpose of mixing other colours? Your question needs some clarification.

Kobi_Z14-Jun-12 7:16

14-Jun-12 7:16

first forgive my poor english, hope i can explain better now.

It is a riddle spread out at my university.
you have 3 basic colors (no other colors are available besides the basic colors and the ones produced by mixing the basic colors)
and a machine which can mix them according to the formula i wrote.

the target color is created by mixing the basic colors or their products in the machine.

some colors take up to 1000 mixing steps so the way i try to "solve" it can take forever.
i dont know much about solving such problems so i ask for help here.
thanks.

Richard MacCutchan14-Jun-12 7:30

14-Jun-12 7:30

Kobi_Z wrote:
It is a riddle spread out at my university.

That makes it clear. However, this being primarily a technical programming forum, I'm not sure if you have come to the right place.

Kobi_Z14-Jun-12 8:07

14-Jun-12 8:07

any suggestions where i can ask ?
on math forum i was told to go to programmers forum...
thanks.

Richard MacCutchan14-Jun-12 21:40

14-Jun-12 21:40

Sorry, no idea.

General App Design Questions

R. Giskard Reventlov14-Jun-12 9:46

R. Giskard Reventlov

14-Jun-12 9:46

This[^] might help.

"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur." Red Adair.
nils illegitimus carborundum

me, me, me

Kevin Marois6-Jun-12 6:02

Re: General App Design Questions

6-Jun-12 6:02

These questions are more about validating how I create apps. I know how I like to do things, but I'm curious how the rest of you do it, so feel free to critique or post suggestions.

I created a playground WPF app for tracking projects. The purpose of this app is primarily to play around and learn. In it I have UI, Entities, DAL, and BL projects. In the DB I created Company, Client, and Project tables. The Projects table has FK's for 3 lookup tables. See the pic here[^]

My project entity looks like this

public class ProjectEntity : _EntityBase
{
    public int ClientId { get; set; }
    public int ProjectTypeId { get; set; }
    public int ProjectStatusId { get; set; }
    public int PayTypeId { get; set; }
    public string ProjectName { get; set; }
    public DateTime? EstStartDate { get; set; }
    public DateTime? ActualStartDate { get; set; }
    public DateTime? EstEndDate { get; set; }
    public DateTime? ActualEndDate { get; set; }
    public bool? IsActive { get; set; }
    public decimal PayRate { get; set; }
}

Ok, now for the questions:

1) Would you have your entity objects maintain references to other entities, or just the PK? For example,
in the ProjectEntity I have ProjectTypeId which stores a reference to the Project Type row in lkpProjectTypes.

When I get a project entity, should I create and store a ProjectType entity on it, or just the PK?

What I usually do is store the PK of the lookup table, then call a method in the DL to get the lookup entity when
it's needed. But I could make the GetProject method in the DL create related object when its called. The
problem is that now you have a heavy object when you may not always need the child objects.

2) Similar to #1. When you create an instance of the ProjectEntity, would you create a ClientEntity and store it
on the ProjectEntity instance? If so, do you then go up the logical tree and create a CompanyEntity to store
on the ClientEntity which is stored on the ProjectEntity? Conversely, if you create a ClientEntity, do you store
a list of ProjectEntities on always? If all I want is to display the Client's name, then getting a fully loaded
ClientEntity seems like overkill.

3) How do you handle nulls, both in the DB and in the app? I'v heard some folks say never store nulls. I don't agree
with that, but I'd like to hear your thoughts.

Thanks

If it's not broken, fix it until it is

Eddy Vluggen6-Jun-12 7:04

Eddy Vluggen

6-Jun-12 7:04

Kevin Marois wrote:
1) Would you have your entity objects maintain references to other entities, or just the PK? For example,
in the ProjectEntity I have ProjectTypeId which stores a reference to the Project Type row in lkpProjectTypes.

When I get a project entity, should I create and store a ProjectType entity on it, or just the PK?

I'm passing the integers around that represent the PK; a list of integers is simple, has little overhead and makes debugging easy. It's also not a large amount of "work" to fetch an ActiveRecord object based on it's handle.

Kevin Marois wrote:
2) Similar to #1. When you create an instance of the ProjectEntity, would you create a ClientEntity and store it on the ProjectEntity instance? If so, do you then go up the logical tree and create a CompanyEntity to store on the ClientEntity which is stored on the ProjectEntity? Conversely, if you create a ClientEntity, do you store a list of ProjectEntities on always? If all I want is to display the Client's name, then getting a fully loaded ClientEntity seems like overkill.

It would add some complexity to ensure that you don't waste too much resources. I do not see many additional benefits to this approach though.

Kevin Marois wrote:
3) How do you handle nulls, both in the DB and in the app? I'v heard some folks say never store nulls. I don't agree with that, but I'd like to hear your thoughts.

I think people should use arguments to defend their one-liners. It is said that a normalized database does not allow for "empty" values; on the other hand, the database-server provides you with the option. In practice, it's easier (and cleaner) to have a nullable field, than it is to have a new set of facts.

Bastard Programmer from Hell Suspicious | :suss:

Re: General App Design Questions

jschell6-Jun-12 9:27

jschell

6-Jun-12 9:27

Kevin Marois wrote:
Would you have your entity objects maintain references to other entities, or
just the PK?

Depends on the nature of the objects and how they are used in the rest of the code.

Kevin Marois wrote:
The
problem is that now you have a heavy object when you may not always need
the child objects.

Yes that is the problem.

First however it must in fact be 'heavy' in terms of the production system.
So 10 rows isn't 'heavy' but one million certainly is. And if a sub entity represents 100 meg of data then even 10 of them is too many. But for the one million case it would always need to do a deferred look up.

For larger ranges with more moderately sized (much smaller sub entities) it depends on actual data and how it will be used.

Kevin Marois wrote:
When you create an instance of the ProjectEntity, would you create a
ClientEntity and store it
on the ProjectEntity instance? If so, do you then
go up the logical tree and create a CompanyEntity to store
on the
ClientEntity which is stored on the ProjectEntity?

That question doesn't make much sense to me. A 'company' seems likes an owner. You can't create an owned object until you create the owner. And you shouldn't even allow for that possibility. Thus in the GUI the user must select a 'company' before doing anything with a owned object.

Kevin Marois wrote:
3) How do you handle nulls, both in the DB and in the app

I handle them as nulls. Presumably for the "app" you mean in the data object. If you mean system wide then the question is wrong because they way a business object deals with nulls is different than how the GUI deals with them.

Application Security

Kevin Marois6-Jun-12 5:56

Encoding and decoding a text file

6-Jun-12 5:56

I'm curious what people are using for application security in WPF/Windows apps? I personally don't like the .Net Framework Security. I have rolled my own user/rights tables and associated code.

If it's not broken, fix it until it is

Re: Application Security

Eddy Vluggen6-Jun-12 7:08

Eddy Vluggen

6-Jun-12 7:08

Kevin Marois wrote:
I'm curious what people are using for application security in WPF/Windows apps? I personally don't like the .Net Framework Security.

I'm relying on Active Directory; has little to do with likings, I'm simply outsourcing that part of the development to Microsoft $Shucks | :-\$

Bastard Programmer from Hell Suspicious | :suss:

Basavaraj Neelagund1-Jun-12 3:54

Basavaraj Neelagund

1-Jun-12 3:54

I am doing a project title "Evaluating the efficacy of FEC coding in Network packet loss" I need report and code for this project..

Richard MacCutchan1-Jun-12 6:26

1-Jun-12 6:26

This is really not a valid question. You cannot just post the title of your thesis and expect someone to provide the complete solution for you. If you cannot make the effort to do your own work then you cannot reasonably expect anyone else to do it for you?

jschell4-Jun-12 8:17

jschell

4-Jun-12 8:17

Basavaraj Neelagund wrote:
need report and code for this project..

Steps to achieve that.
1. Research the topic.
2. Learn to code
3. Write the code
4. test the code
5. Write the report (presuming that is different than the code.)

Albert Holguin4-Jun-12 8:39

Albert Holguin

4-Jun-12 8:39

FEC in network packet loss? ...packets use CRC checks to determine errors, then retransmit when detected... are you sure your report even makes sense? D'Oh! | :doh:

R. Giskard Reventlov4-Jun-12 8:44

R. Giskard Reventlov

4-Jun-12 8:44

and I need money; you give me money and I'll tell you where to find what you want. Smile | :)

"If you think it's expensive to hire a professional to do the job, wait until you hire an amateur." Red Adair.
nils illegitimus carborundum

me, me, me