|
In the past I worked in a team of 100 developers on a very big project. The rule was to build everything in the house and avoid "out of the box" solutions. But we where 100 developers, we had the budget, the time and that was the good period and there wasn't a lot of good "out of the box" solutions.
After that I worked in smallest team with people pretending working on big project and it's probably true from their point of view but for me, because of my background, it was small teams and small projects. according to my experience in smallest team developers prefer to work with many "out of the box" solutions. Most of the time they don't have the budget or time to build their own solutions.
I insisted on the "pretending working on big project" because I often read information about people saying they use a lot of "out of the box" solutions in their "big" team or in their previous experience. But I can't imagine how they do when they have to manage their team, new developers and competencies, branching management with 5 or more versions at the same time, and trying to maintain all version of all your thirds party tools. As you said it's impossible to maintain a big project when you have to care about dependencies and updates of these third party frameworks.
I personally divide the developer in two categories just for fun. The alchemist and the magician. The first group are the developers that prefer to compose their code with "out of the box" solutions. The second group are the developers that prefer to build themselves everything. Of course most of people are somewhere between these two extreme.
EDIT -- I just learn that the term of Voodoo programmer already exists. My Alchemists are these Voodoo programmers.
modified 20-Jun-14 2:54am.
|
|
|
|
|
Hey guys, thanks for checking my post out. I'm writing this in hopes of getting some general feedback. Particularly, on my use of the design pattern and the abstractions I came up with.
I've been coding on-and-off for sometime now, but I never thought about design patterns until recently. I'm realizing how much I've missed out on  ! But, anyway, I came across the decorator pattern in Head First Design Patterns and it got me thinking about a problem I couldn't solve a few years back...
It happened that I needed to enumerate all the different ways to choose k things from n things. The formula from discrete math tells you the total number of different ways to choose, but I needed to know what they were. I came up with a solution easily enough, but it was not acceptable. My solution (listed below) printed out all the different ways to choose 3 things from 6 things.
for(int item1 = 0; item1 < 6; ++item1){
for(int item2 = item1+1; item2 < 6; ++item2){
for(int item3 = item2+1; item3 < 6; ++item3){
System.out.println(item1 + " " + item2 + " " + item3);
}
}
}
Hopefully, it's obvious why it sucked! I could not vary, at run-time, the number of items I was choosing  . I could only vary the total number of things to choose from. Choosing an additional item required hard-coding another nested loop. So, I caved in and just restricted the user to a range and coded for every case  .
It was when I was learning the decorator pattern that it hit me. They kept talking about how decorators 'wrap' and 'contain' things and how they can be built dynamically. The figure[^] they used to illustrate the concept screamed "NESTED FOR-LOOPS" at me. I HAD to try and apply this design pattern to my problem. It didn't matter if it did not work or if there was a better solution; I'd count it as practice.
I'm going to assume you are familiar with the Decorator Pattern[^], but I provided the link if not. So, basically, my abstract component is ItemSet and my only concrete component is ChooseItem. The abstract decorator is AdditionalItems and the concrete decorator is ChooseAdditionalItem. Hopefully, the way I named them makes it easy to picture how they work together. I have to say, IMO its pretty nice how effortless the solution looks in the main function. I've written a sample below.
int setSize = 6;
int numberOfItemsToChoose = 3;
ItemSet itemSet = new ChooseItem(setSize);
for(int i=0; i < numberOfItemsToChoose-1; ++i)
itemSet = new ChooseAdditionalItem(itemSet);
Queue<String> waysToChoose = itemSet.chooseItems();
for(String aWayToChoose : waysToChoose)
System.out.println(aWayToChoose);
I promise the rest of the code is very short and human-readable, but I'd really, rather not post all the code here. Instead I'm providing a link[^] to it.
So, if you took a look at the code - thank you if you did! - you might have seen that this new solution is recursive. Meaning that all the different ways to choose are held in memory before they actually get printed out, whereas in the first one they were printed out immediately.
Any kind of feedback is appreciated. If you know of the best solution for this let me know or if you see a non-recursive solution, let me know too. Also, I am unemployed! So, help with that too would be amazing!
modified 5-Jun-14 23:48pm.
|
|
|
|
|
I am working on an embedded project with 2 microcontrollers, uC A and uC B. They are linked by an SPI-bus and uC A is the master and uC B is the slave. Both of the microcontrollers have read/write registers and uC A should be able to read and modify uC B's registers and vice versa. Is there a standard protocol available for this or do I need invent something myself?
|
|
|
|
|
|
I want simple numerical concepts related to Registers, Data Bus and Address Bus.
Please can any one give a clear concepts of solving numericals related to Registers, Data Bus and Address Bus.
example problems like:
If a memory had 16 bit address bus and a 32 bit data bus what is the largest size of the memory?
This will be very beneficial for me.
Thanks!
|
|
|
|
|
nischalinn wrote: This will be very beneficial for me.
Learning to do this[^] will be even more beneficial.
|
|
|
|
|
Hi, first of all, sorry if you don't understand something, english is not my main language.
I have to make a mini bank system (it's an assignment), but I don't really know how banks work. I'm thinking of making to web systems, one for the client and another one for the bank agent, but I don't know if there's a better way to do it. I would add a third kind of account, which is the Admin, to administrate the bank agent accounts and stuff, but I'm kind of lost on how banks actually work with this. I also need to withdraw (I don't know if this word is correct) money from an account, so I will need another app for the ATM (I can just think of the ATM as a "smart-atm", with touch screen and stuff, not necessarily like the ugly ones that exist now). Is it really necessary that I create 3 different systems? can you help me on the "design" of the problem?
I translated the basic requirements:
A bank system is needed. This system must allow deposits on a bank account. Deposits can be made in cash or by check. The system must keep a log of all the movements in the accounts. A "bank agent" can create client accounts, and he will have access to see their movements information, like amount, date, hour, and destination (transfers, deposits, etc). The application must allow:
Deposits by check.
- Validate the info. (ie: serial of the check)
Deposits in cash.
- Validate the info: client info, account number, etc.
User account management
- Profiles: Bank agent, client.
Bank account management
- User "Bank agent" can freeze bank accounts
- User "Client" can see the movements of his account (by date ranges, or a particular day, etc)
The system must have a service to withdraw money from the bank account
Thanks.
|
|
|
|
|
Do you have any restriction on the technology to be used?
|
|
|
|
|
Oh yes, sorry, I forgot. I have to use Java EE and Mysql server.
|
|
|
|
|
I would choose just one application with three different profiles.
Would opt for a presentation layer, a layer with the domain model or business layer and a data layer.
The presentation layer would then be a web application, the business layer would contain all the logic and finally the data layer that abstracts the business layer operations with the database.
The business layer in my opinion should be modeled using the domain language.
Do you already create your data model?
|
|
|
|
|
sebasuraa wrote: but I don't really know how banks work
You are confusing things.
You don't need to know how banks work.
What you need to know is how to implement the assignment that you were given. This should be based on something that you have already done.
Presumably you have already been introduced to designs or perhaps Agile. So that is where you start. Although if not you can still start by just coding one single feature.
|
|
|
|
|
I am building a sharepoint portal for providing services for customers.
the customer uses this portal to submit request that includes his data and his attachments. The attachments will be uploaded in a document library and the data will saved in a sharepoint list.
the expected count of requests that will be provided yearly is 5000 requests, and every request includes 10 attachment files.
What is the best approach that I should follow??
1- using SharePoint list for saving the requests in & using document library for saving the attachments.
2- using list & document library. and make folder for each year in the list and document library.
3- using database table for saving the data and using the document center in sharepoint for saving the attachments.
4- there are another approaches for this user requirement .
Thanks
Maher Elsayed
|
|
|
|
|
I was wondering if it is or is not. I know i586/686 is patched to all new software.
|
|
|
|
|
By who?
Windows won't run on it (except versions like win95 and NT3.5) so whether windows software is still compiled with 386 compatibility (it isn't) is kind of a moot question.
Even Linux doesn't support it anymore, as of 3.8. The demand for it dropped to zero a decade ago, and the support required a significant amount of "stuff that's only there because of a processor that no one uses".Wojciech Lisik wrote: I know i586/686 is patched to all new software.
That isn't really true either. There is absolutely software out there that doesn't have a fallback in case SSE2 isn't supported, and it's often (incorrectly) still called "i686 compatible". Requiring higher versions of SSE is becoming a thing too, and it's only going to get more widespread.
|
|
|
|
|
Maybe as an embedded CPU?
|
|
|
|
|
Guys,
I have a scenario which need your advice on it.
I try to keep it short here.. So not putting whole thing in.
I need to develop a system which require approval to maintain any record.
Below is my DB design
Student Table
- StuId
- StuName
- StuAddress
- RecStatus (Deleted, Active, Pending Add, Pending Edit or Pending Delete)
Approval Table
- AprId
- AprCat (Add Request, Edit Request or Delete Request)
- AprModule (Which table to refer to)
- AprModuleId (Target Record)
- AprStatus (Pending Approval, Approve, Reject)
- AprBy
- AprDT
Approval Detail Table (Handle Edit Request Record)
- AprDId
- AprId
- UpdField (Which Field to update)
- NewValue (New Record)
For Adding record, system will insert the record into the student table and have it update as Pending Add in RecStatus. The Id will then added to Approval table for Approval Request.
For Delete Record, system will update the record as Pending Delete in RecStatus. The StuId will added to approval table for approval request.
For Edit record, system will not change the record in student table. But it added into the Approval Detail table and sit there for approval. If request approved, system will loop thru the approval detail record and update the fields one by one.
This is how I do it,
Was thinking is there any better way to work with this kind of scenario?
If yes, please point me to the right direction.
Thank
CA
|
|
|
|
|
Based on the information provided, I think you already have the best idea so far. Well done!
|
|
|
|
|
|
I disagree with the entire design concept of delaying a change until it is approved, it feels like a nightmare waiting to happen, the first time someone needs an "urgent" change implemented the workflow will be short circuited by the needs of the business to get the job done.
I use a logging system (almost all system rely on logging) make the changes immediately and record the information in a log file, if there is a problem you can trace the perpetrator and reverse the changes if required. This allows the business to do their job quickly and gives you the recovery information when they screw up.
Never underestimate the power of human stupidity
RAH
|
|
|
|
|
Hi, I like your quote "Never underestimate the power of human stupidity".
The concept is requested by the user which what they been practice right now. No matter how urgent the changes is, they need to go thru the approval before getting the changes out for others to use it.
We do have a log system which log all changes and approval in the system. I agree will your suggestion, as the whole process will be very quick and no delay. But this process will be great if all user know that what is a correct record looks like. Most of the user will just use whatever data display in the system to get the job done. And it will be a mess when 'Someone' found out is a wrong data. It will take a lot of money and resources to reserve it(In Business).
For us, no record is better than wrong record.
Thank for the sharing. 
|
|
|
|
|
Ah user requirements, they will drive you nuts at some stage.
I can understand the requirement (I thought the student thing was only an example) I do suggest you put in place as much validation as you can think of to reduce the need for intelligent input. You may end up with only a few critical fields that need autorisation.
Never underestimate the power of human stupidity
RAH
|
|
|
|
|
Ya, they are driving me nuts now..
Yes, the student thing is an example but the process is not.
Will try to think off any other validation required like what you mention.
Thank for the suggestion.
|
|
|
|
|
Cee Ann wrote: Was thinking is there any better way to work with this kind of scenario?
Ehr.. how does a multi-table update look like (in a transaction please) with someone "approving" each record before it's edited? And what are you going to do with concurrent edits?
Say both I and your boss edit the same record. Now both edits are awaiting approval. And both of us need the ID that's generated, so we can enlist the student into a class. Whose edit will win? It better be my edit 
Whoever felt the need to give his permission; keep him away from the database-design. There's a GOOD reason why a developer creates the design, and not the user. His permission might be required for a business-rule, but it has nothing to do with how we physically store data.
Aw, and why are you prefixing your fieldnames? And why is it already inconsistent? Wouldn't it be a bit more readable if you omit the prefix?
Bastard Programmer from Hell 
If you can't read my code, try converting it here[^]
|
|
|
|
|
Hi,
For multiple table update, are you referring to other table such as classes or course?
On concurrent edit, we only allow record in active status to be edit. If the record is in Pending Edit Approval Status, the record will not be allow to edit again. Until the pending request is approved/rejected. This is been taken care in the RecStatus field of each table.
So, it will not happen that you and my boss edit the same record in same time.
But If it really happen, I will let me boss win. As he is paying my salary.
The database is design by me. If you see any issue with it, please point it out. So I can learn from my mistake.
Forgive my laziness on the prefixing of fieldname, It all started when I writing my SQL statement and felt it's too long.. So I try to shorten it a little bit.
Was learning my way to not prefixing my fieldname with EF6, as I get the suggestion function in VS.
Thanks for the update.
CA
|
|
|
|
|
Cee Ann wrote: For multiple table update, are you referring to other table such as classes or course?
I don't have the entire schema so I can't say to which tables it applies. Let's take a fictive person called Jane and enlist her into three courses. That's 4 items that need be "approved".
Now enlist 400 students in 40 different courses
Cee Ann wrote: If you see any issue with it, please point it out
Whether it is valid depends on the users' needs; I can't tell whether it's valid, but one can always apply the normalization principles. To me it'd be valid if it is in BNF.
Cee Ann wrote: Forgive my laziness on the prefixing of fieldname
Lazy? I was wondering why you're doing all that work, when it doesn't seem to help much
Bastard Programmer from Hell
If you can't read my code, try converting it here[^]
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
![Rose | [Rose]](https://codeproject.freetls.fastly.net/script/Forums/Images/rose.gif)
