|
Thread Rules 1. This is not a "do my homework for me" thread. If you have specific questions, ask, but don't post an assignment or homework problem and expect an exact solution. 2. No recruiting for your cockamamie projects (you won't replace facebook with 3 dudes you found on the internet and $20) 3. If you can't articulate why a language is bad, don't start slinging shit about it. Just remember that nothing is worse than making CSS IE6 compatible. 4. Use [code] tags to format code blocks. |
On May 27 2015 16:19 netherh wrote:Show nested quote +On May 27 2015 06:40 Nesserev wrote: ...
for (auto& pair : myMapB) myMapA[pair.first] = myMapA[pair.first] + myMapB[pair.first];
...
You can probably shorten that a bit:
for (auto const& pair : myMapB) myMapA[pair.first] += pair.second;
(You're not changing "pair", so use const& instead of &. No need to do the myMapB lookup, since the value is in pair. No need to do two lookups in myMapA (though I suspect that might get optimized anyway)).  I'd be up for collaborating on something, if the project is interesting.
Thank both of you, I figured out my problem got solved by me just adding a getMap-function, which I didnt do initially for some reason. I guess I should take more breaks while coding rather than working on a problem non-stop. The finaly solution already existed in my code, but commented out since I couldnt make it work without getMap.
A few question regarding the provided code though: auto& pair - the & is used to use the adress of pair, rather than copying it, which our memory apreciates.
const - indicates that pair doesnt get changed. I assume this is for improved readability of the code aswell as security. Does it also increases performance, decreases performance due to having to check if pair didnt get changed or does it have no effect on the performance at all?
|
On May 27 2015 21:03 Artesimo wrote: const - indicates that pair doesnt get changed. I assume this is for improved readability of the code aswell as security. Does it also increases performance, decreases performance due to having to check if pair didnt get changed or does it have no effect on the performance at all?
No effect on performance at all: "const" is only verified at compile time unless I am mistaken.
|
Compilers -may- be able to optimize a variable declared const in some cases. However, it's really quite a micro-optimization.
|
On May 24 2015 02:33 Morfildur wrote: I'm developing the best PHP Framework that ever walked the tubes of the internet because I'm tired of the crappy PHP Frameworks that are available. I've tried most of them, basically all of the big ones, and all of them are just terrible, so I'm doing it right now with proper view models, hierarchical structure, full separation of views, controllers and models, not the pseudo-separation that all the frameworks seem to like, fully configurable, ... Of course it's a long way away from perfection, especially considering that I'm currently in the third rewrite of almost the whole structure because I had a great idea that didn't fit into the existing structure. That's cool but it suggests your structures aren't very flexible.....might want to focus on improving that.
|
I am stuck again. I try to replace my set/get methods with the overloaded [] operator and my function that megres two maps with the + operator. However, I am horribly stuck again. I have multiple entitys of abstract datatype, each of them storing an object that contains a map.
The abstract datatype is SparseMatrix and contains a pointer to one object MatrixElement. MatrixElement contains std::map.
I already figured out that
Stash operator+(const Stash& stashB) { Stash stash; stash.item1 = this->item1 + stashB.item1; stash.item2 = this->item2 + stashB.item2; stash.item3 = this->item3 + stashB.item3; stash.item4 = this->item4 + stashB.item4; return stash; }
stashA= stashB + stashC;
works well if I do all off this inside my main, I struggle however with doing so in my case since I have a hierachy that goes like: [node] - holds my "SparseMatrix"s in a std::vector. then [SparseMatrix] - each holding one"MatrixElement" then [MatrixElement] - each holding one map.
I cant figure out where and how to implement the overloading. I thought it had to be in [node] since it is the very top, but then I struggled since this-> accesses [node] and I only habe one pargument that I can use to adress which element of vector<SparseMatrix> I want to adress, I would need 2 though.
The []-operator is even worse, I thought it would be easyer, but I fail at getting my map-layout to work/make it work at all.
std::map<std::pair<int, int>, int>
int MatrixElement::operator[](std::pair<int, int>) { //dont know how to make the function accept variables, since // return map[int, int]; // return map[std::pair<int, int>]; // return map[std::make_pair(int int)] // dont work. }
|
On May 27 2015 23:34 fmod wrote: Compilers -may- be able to optimize a variable declared const in some cases. However, it's really quite a micro-optimization.
Indeed it is a "micro-optimization". + Show Spoiler +peephole optimization is a slightly more correct phrase that means the compiler is looking at the code through a peephole and optimizing only a few nearby lines at a time.
GCC and most other modern compilers use the very effective constant folding optimization. This ends up being a huge code-size save overall. Adding a single instance of const may only save a few instructions, but it can also allow the compiler to do other optimizations on future passes and greatly reduce the code.
One thing to mention is that the compiler is actually pretty smart about knowing which variables will change during execution, so it will pepper in it's own const modifiers in order to do constant folding. This optimization is turned on by default.
|
On May 27 2015 09:24 Blitzkrieg0 wrote: What are you planning exactly? The language is irrelevant until you have an idea what is being built.
Nothing in specific, but I was hoping for us as a community to come up with a nice idea.
|
On May 28 2015 04:43 darkness wrote:Show nested quote +On May 27 2015 09:24 Blitzkrieg0 wrote: What are you planning exactly? The language is irrelevant until you have an idea what is being built. Nothing in specific, but I was hoping for us as a community to come up with a nice idea.  I wouldn't...mind you guys working on PvPGN
|
On May 24 2015 01:56 darkness wrote: What projects do you guys do outside university and work? Every time I have some time to code something, I don't have an idea for a personal project. I don't do programming exercises as they're boring, but I can't come up with something practical that I need to use/automate.
I've actually been meaning to share my own project:
Phamda is functional programming library for PHP. The main goal is automatic partial application or currying, so you can for example do this:
use Phamda\Phamda as P;
// Add function is roughly equivalent to basic addition operator: P::add(15, 27); // => 42
// But it is also has automatic partial application: $addFour = P::add(4); // Create a new function $addFour(5); // => 9 $addFour(16); // => 20
Placeholder arguments supported as well:
$div15 = P::divide(15); // Create a new function that uses 15 as dividend $div15(3); // => 5
$_ = P::_(); // Create / retrieve a singleton placeholder object
$divBy5 = P::divide($_, 5); // Create a new function that uses 5 as divisor $divBy5(20); // => 4
About 100 functions are available currently, including basic functional tools and functions for manipulating lists, collections, strings and objects. The functions are easily composable, allowing the creation of more complex functions and function pipelines. More examples can be found on GitHub or in the documentation.
Hopefully someone will find this useful. Any comments and questions are of course welcome -- here or on GitHub!
|
On May 28 2015 01:43 Artesimo wrote:The []-operator is even worse, I thought it would be easyer, but I fail at getting my map-layout to work/make it work at all. std::map<std::pair<int, int>, int>
int MatrixElement::operator[](std::pair<int, int>) { //dont know how to make the function accept variables, since // return map[int, int]; // return map[std::pair<int, int>]; // return map[std::make_pair(int int)] // dont work. }
It would be helpful to know what "don't work" means. What should this do? What does it actually do? (compiler error, crash, wrong behaviour?)
I'm guessing that you just want to wrap std::map's operator[]. In which case you'd want something like this:
int& MatrixElement::operator[](std::pair<int, int> const& key) { return map[key]; }
Returning an int& means that you can change the actual element inside the map. (e.g. matrixElement[{1, 2}] = 5;) If you return by value, you're copying the int. So you're not changing the actual map at all.
EDIT: for your hierarchy, try thinking about what functionality each level adds, and the interface it needs to provide to its users. It kinda sounds like you have [node [sparse matrix [matrix element] ] ], with no functionality added at each level. In which case I'd suggest merging them all into one SparseMatrix class.
If this isn't the case though, think about what you want to do with each level. Try writing some fake "user" code that uses each class's interface.
|
UPDATE table SET column = column * 10 WHERE foreign_id < 123123
Error: #1064 - You have an error in your SQL syntax; check the manual that corresponds to your MySQL server version for the right syntax to use near 'FROM table WHERE foreign_id < 123123' at line 2
WTF? Stupid phpmyadmin.
|
Anyone that got a good tutorial/online reference on how a Red-black tree works? I can't wrap my head around when a node is supposed to change colour from red to black or vice versa. Also I don't really get why the root and nil leafs are black, or why that both the children of a red node are black, but a black node can have black child nodes.
|
On May 28 2015 22:44 Kamari wrote: Anyone that got a good tutorial/online reference on how a Red-black tree works? I can't wrap my head around when a node is supposed to change colour from red to black or vice versa. Also I don't really get why the root and nil leafs are black, or why that both the children of a red node are black, but a black node can have black child nodes.
A node(s) changes color when the structure is no longer a red-black tree (insertion or deletion). When you insert the node is always black. Then you need to check the path to the root to make sure the red-black structure is maintained, fixing the tree as you head up to the root. Deletions are a bit more complicated, but you can find an animation of them on youtube by searching red-black tree animation.
A black node can have black children because the structure wouldn't work otherwise. If you merely alternated then the red-black tree wouldn't mean anything. The root has to be black because a two node tree can't exist with a red root node. The null leaves being black just increases the entire trees black height by one so you can not worry about them at all if it confuses you. Just draw your tree without the null pointers.
|
On May 24 2015 02:33 Morfildur wrote:Show nested quote +On May 24 2015 01:56 darkness wrote: What projects do you guys do outside university and work? Every time I have some time to code something, I don't have an idea for a personal project. I don't do programming exercises as they're boring, but I can't come up with something practical that I need to use/automate. I'm developing the best PHP Framework that ever walked the tubes of the internet because I'm tired of the crappy PHP Frameworks that are available. I've tried most of them, basically all of the big ones, and all of them are just terrible, so I'm doing it right now with proper view models, hierarchical structure, full separation of views, controllers and models, not the pseudo-separation that all the frameworks seem to like, fully configurable, ... Of course it's a long way away from perfection, especially considering that I'm currently in the third rewrite of almost the whole structure because I had a great idea that didn't fit into the existing structure. Though... I think I spent more work time on it every time I was bored or finished a task faster than expected than actual time out of work.
Lol, everyone mentions they're making something better than the other and end quitting half way through because of the amount of time it requires. Have a git?
|
Hyrule18977 Posts
|
On May 28 2015 17:10 netherh wrote: int& MatrixElement::operator[](std::pair<int, int> const& key) { return map[key]; }
Returning an int& means that you can change the actual element inside the map. (e.g. matrixElement[{1, 2}] = 5;) If you return by value, you're copying the int. So you're not changing the actual map at all. EDIT: for your hierarchy, try thinking about what functionality each level adds, and the interface it needs to provide to its users. It kinda sounds like you have [node [sparse matrix [matrix element] ] ], with no functionality added at each level. In which case I'd suggest merging them all into one SparseMatrix class. If this isn't the case though, think about what you want to do with each level. Try writing some fake "user" code that uses each class's interface.
Thank you for your help. For the hierarchy: I already figured out that I dont need that much classes, but my prof wants it that way, so I just have to deal with it.
For the []-operator, your solution isnt working for me somehow. From my tutor I got the hint that it would be good if the end result would work like MatrixElement[x][y]=z and std::cout<<MatrixElement[x][y]
I figured out I might need another class, so I can overload the operator of MatrixElement and MyOtherClass. MatrixElement[x] then returns something that [y] can work with(like returning an object of MyOtherClass), since it seems impossible to overload [] so that it accepts multiple arguments on its own. For that, MyOtherClass needs access to MatirxElement's map, since it has to return the value for map[x,y].
So far I only managed to overload [] for matrixElement, working fine if I have a map with a single key. As soon as I add MyOtherClass to manage MatrixElement[x][y], I get horribly stuck and now I just have a huge mess of code where I dodnt even find my initial attempt in thorugh the sheer amount of code that I commented out because it wasnt working/I got an idea that I wanted to follow through after lol.
|
On May 29 2015 00:33 ShoCkeyy wrote:Show nested quote +On May 24 2015 02:33 Morfildur wrote:On May 24 2015 01:56 darkness wrote: What projects do you guys do outside university and work? Every time I have some time to code something, I don't have an idea for a personal project. I don't do programming exercises as they're boring, but I can't come up with something practical that I need to use/automate. I'm developing the best PHP Framework that ever walked the tubes of the internet because I'm tired of the crappy PHP Frameworks that are available. I've tried most of them, basically all of the big ones, and all of them are just terrible, so I'm doing it right now with proper view models, hierarchical structure, full separation of views, controllers and models, not the pseudo-separation that all the frameworks seem to like, fully configurable, ... Of course it's a long way away from perfection, especially considering that I'm currently in the third rewrite of almost the whole structure because I had a great idea that didn't fit into the existing structure. Though... I think I spent more work time on it every time I was bored or finished a task faster than expected than actual time out of work. Lol, everyone mentions they're making something better than the other and end quitting half way through because of the amount of time it requires. Have a git?
I have a .tgz of the project I've started a few days ago and spent maybe an hour on here. The git archive is protected by ssh keys on my private server and I can't be bothered setting up public access.
I haven't completely quit yet. I'm currently using it for the third application to see how it works for that use case (Basic weapon review/voting website for Planetside 2) and finding some more things to tweak a bit here and there. Previously I used it for a webbased game I abandoned and the Planetside 2 outfit (guild) forum.
I'm not too happy about the Domain Models (e.g. Application/PS2Weapons/Domain/Model/Vote.php) atm since the auto-generation I used beforehand just wasn't flexible enough, so I'm now writing them by hand again. They are very simple because I have everything in stored procedures in the MySQL backend, e.g. creating a forum post is just $dbh->prepare("CALL AddPost(:user_id, :thread_id, :content)") though I still think I could simplify them more. However, every time I make them flexible the description of the domain model ends up being just as big and tedious as writing the actual model itself.
Once I've decided on that, I need to clean up the directory structure, move Twig out of the Framework directory, remove some old application specific data out of that folder as well (Markdown) and all of that. After that it's time to clean up the bootstrap and make the services dynamic instead of using a fixed list of services and all of that, make the databases dynamically created on use rather fixed at the start, ...
I also haven't upgraded the user model to the newest structure where most of the controllers are no longer necessary and instead replaced by a configuration-based combination of generic controllers, views and viewmodels.
Oh, and I broke caching and removed it instead of fixing it until I have the rest done. It's fast enough anyways.
Some basic rules I use: - static is evil and only used where absolutely necessary. I have one place where I still use static, which I want to get rid off but haven't yet gotten around to. - All data handling should be done inside the database itself, including joins and such that are only exposed via views while data manipulation is only allowed through stored procedures. Every query that might be slow will be sped up by using trigger or SQL procedure based denormalization to remove on-the-fly calculations (e.g. number of votes is increased inside the AddVote() SQL Procedure). - As much as possible should be done at deploy (compile) time, that means all javascripts get compiled into one, the module configurations get parsed and created for the environment (dev/stage/live), so reading the actual configuration is done directly without any application logic to figure out which environment we're in.
So much to do, so little time and so many more ideas...
It's a living project and still very messy, but every time I'm bored at work, it gets a little bit better (or a little more broken). It's still fun working on it even though I will be the only one to ever use it.
|
On May 29 2015 02:35 Artesimo wrote: I figured out I might need another class, so I can overload the operator of MatrixElement and MyOtherClass. MatrixElement[x] then returns something that [y] can work with(like returning an object of MyOtherClass), since it seems impossible to overload [] so that it accepts multiple arguments on its own. For that, MyOtherClass needs access to MatirxElement's map, since it has to return the value for map[x,y].
So far I only managed to overload [] for matrixElement, working fine if I have a map with a single key. As soon as I add MyOtherClass to manage MatrixElement[x][y], I get horribly stuck and now I just have a huge mess of code where I dodnt even find my initial attempt in thorugh the sheer amount of code that I commented out because it wasnt working/I got an idea that I wanted to follow through after lol.
Yeah. You're on the right track.
You either have to go for [x][y], which means returning a proxy class, or (what I'd recommend) provide an access function with a different name (.get(x, y) or perhaps better .getOrDefault(x, y)). (Some people might overload operator()(x, y), but I think this is less clear).
If you do go for the proxy class, you'd need something like:
struct Proxy { explicit Proxy(MatrixElement& element, int x): element(element), x(x) { } int& operator[](int y) { return element.map[std::make_pair(x, y)]; } private: MatrixElement& element; int x; };
Proxy MatrixElement::operator[](int x) { return Proxy(*this, x); }
Proxy will need to be a friend of MatrixElement (or use a .getOrDefault(x, y) member of MatrixElement itself). You'd also need a separate proxy object for const access.
So... it's much simpler and clearer to just provide access using a different function than operator[]. And you can name it something more helpful too.
|
I have been hired by a consultant company as "junior consultant". This work will mostly consist of delivering code/programs to Banks/insurances etc. They told me most of it will be in a "object oriented language" like C++ or Java.
I'm a physicist and last time I programmed was in C (and Root, but better not speak of that), mostly simulation stuff, but nothing "big"
Since they have already told me that I will require to use multiple languages I'm unsure what I should concentrate on: C would build foundations and I already have experience with it, the other two are closer to the actual work.
What would you learn and which books would you recommend? (would they be the same if you had just one month?)
PS: I'm absolutely aware that one month isn't enough to make me a programmer.
|
If there's a focus on OOP then you definitely should start learning an OOP language. C++ is hard (as in: you need a very precise understanding of what you are doing and a lot of discipline to do things properly); one month most likely isn't enough for you to become solid at it. But Java has its own share of problems, so pick your poison.
Also if you pick C++, definitely use books to learn. The internet is full of really awful C++ guides. I personally like Stroustrup's "The C++ Programming Language", but I didn't read other books on C++ specifically (I fell into the internet guide trap and dug myself out the hard way). In my opinion one of the most important things to understand in C++ is RAII (resource allocation is initialization), which is a structured approach to memory management. Once you understand and apply RAII properly the majority of your memory management issues will go away. The next important thing is the standard library.
I'd guess you're better off if you focus on one of the two languages. And upon further consideration I'd tend towards suggesting Java. But I have no idea which one would be more important in your actual case.
|
|
|
|