PA Programming Thread :: PAdev.net - Need hosting for a service of yours? Check it out.

bowen

Last I heard they haven't done much with it. Would be fun to dick around with it. I think you're better off getting better results doing it yourself in C and then writing a C# parser of sorts and going from there. It'd probably get more done than any OS project.

Joe K

bowen wrote:

Phyphor wrote:

In my experience pointers confused the shit out of most people

It's surprising really, that someone that's in comp-sci can be confused by the concept of what pointers are. The concept isn't sufficiently difficult but there are just some people that can't make sense of it.

Then there's the difference between reference, pointer, and all that shit that just adds fuel to the fire.

pointers and references should be pretty par for the course for any decent compsci program.

where the real shit gets confusing is STL and Vectors.

bowen

Joe K wrote:

bowen wrote:

Phyphor wrote:

In my experience pointers confused the shit out of most people

It's surprising really, that someone that's in comp-sci can be confused by the concept of what pointers are. The concept isn't sufficiently difficult but there are just some people that can't make sense of it.

Then there's the difference between reference, pointer, and all that shit that just adds fuel to the fire.

pointers and references should be pretty par for the course for any decent compsci program.

where the real shit gets confusing is STL and Vectors.

I don't know, that stuff was comparatively easy. I'm not saying pointers and references are hard, but, the concepts behind them can confuse the fuck out of people.

Nightslyr

I just came across the idea of using output buffering to increase PHP script speed. Aside from the implicit buffering that Apache does, am I wrong in thinking that manually invoking the ob_* functions would be pointless in a MVC environment? I mean, by the time output is supposed to be rendered, all of the heavy lifting is done, with only simple view logic remaining.

Infidel

Yeah, that won't really affect performance. What those functions are good for performance wise is capturing the output so that you can implement caching of it, so that the heavy lifting is not done on every request, up-front is fine but it doesn't persist.

Linden

Jasconius wrote:

Anyone here ever dick around with stuff like Phantom OS or http://en.wikipedia.org/wiki/Microsoft_Singularity ?

This kind of stuff is fascinating to me. I think it would be so cool to have a whole new technology stack based on something other than UNIX.

You might want to read about Synthesis as well. (Massalin, 1992)

Apothe0sis

Familiarising myself with Python and loving it...

Is there a way to call a specific attribute (this is the wrong term, isn't it?) or method from a variable like so:

Like,

>>> b.foo
"bar"

Can you do something like

>>> a = "foo"
>>> b.$a
"bar"

or something so that a is interpreted as an attribute or method name?

I've been trying to google it or read through the language reference, but I'm not sure exactly what this would be called to check into it more efficiently.

Lux782

So, talking unit testing...

How would you unit test the controls of a video game? Things like when A is pressed something happens? I was thinking of a class that you supply the key press and it alters an acceleration vector. Then I can test when a key is pressed the acceleration vector changes to what I expect it to be. However testing things like game logic might be more difficult. For collision detection I could setup a bunch of potential situations and determine if the objects collide. Anyone have any thoughts?

Seol

Lux782 wrote:

How would you unit test the controls of a video game? Things like when A is pressed something happens? I was thinking of a class that you supply the key press and it alters an acceleration vector. Then I can test when a key is pressed the acceleration vector changes to what I expect it to be. However testing things like game logic might be more difficult. For collision detection I could setup a bunch of potential situations and determine if the objects collide. Anyone have any thoughts?

Firstly, to unit test, make sure you're testing units. So if pressing A makes your character jump, then test that when your control class is given that input, that it invokes the jump() method on a mocked character. Then, separately, check that when the jump() method is invoked on your character, that the appropriate accelerations are applied.

For collision detection, yeah: firstly check your collision classes by setting up various situations and seeing if they collide, then test your broad-phase algorithms by setting up mocks

If you don't know how to unit test something because it's too big, or because there are too many steps between input and expected behaviour, that's a sign that you need smaller tests.

Seol

Doublepost

Jimmy King

Apothe0sis wrote:

Familiarising myself with Python and loving it...

Is there a way to call a specific attribute (this is the wrong term, isn't it?) or method from a variable like so:

Like,

>>> b.foo
"bar"

Can you do something like

>>> a = "foo"
>>> b.$a
"bar"

or something so that a is interpreted as an attribute or method name?

I've been trying to google it or read through the language reference, but I'm not sure exactly what this would be called to check into it more efficiently.

This seems to be how to do it here. It's not quite as awesome of a solution as I'd expect from Python, but I guess it'll work.

Looks like you can use eval() as well, but the above solution is safer.

Starfuck

So yeah, diving into native node.js extensions and v8 engine is maybe a bit deep for a c++ novice. Lots of room to learn, but man, sometimes my eyes just glaze over looking over some of this source code.

Apothe0sis

Jimmy King wrote:

Apothe0sis wrote:

Familiarising myself with Python and loving it...

Is there a way to call a specific attribute (this is the wrong term, isn't it?) or method from a variable like so:

Like,

>>> b.foo
"bar"

Can you do something like

>>> a = "foo"
>>> b.$a
"bar"

or something so that a is interpreted as an attribute or method name?

I've been trying to google it or read through the language reference, but I'm not sure exactly what this would be called to check into it more efficiently.

This seems to be how to do it here. It's not quite as awesome of a solution as I'd expect from Python, but I guess it'll work.

Looks like you can use eval() as well, but the above solution is safer.

Looks preeeeetty good - certainly solves my problem. Thanks Jimmy!

The Anonymous

Jimmy King wrote:

Looks like you can use eval() as well, but the above solution is safer.

Pfft. Who needs code safety when you can have Python code in your Python code?

Jimmy King

The Anonymous wrote:

Jimmy King wrote:

Looks like you can use eval() as well, but the above solution is safer.

Pfft. Who needs code safety when you can have Python code in your Python code?

I herd you liek... I can't do it.

The Anonymous

Jimmy King wrote:

The Anonymous wrote:

Jimmy King wrote:

Looks like you can use eval() as well, but the above solution is safer.

Pfft. Who needs code safety when you can have Python code in your Python code?

I herd you liek... I can't do it.

mr_mich

So, my JSP/servlet class is on headers and stuff now. I understand all that, but our prof makes a note that since servlets are multithreaded, we need to be careful of how we use our counters.

Am I to assume I need to use a synchronize{} block to make my stuff thread-safe? Something like this:

int pageViews;

public void doGet(...){
  synchronize(pageViews){
     pageViews++;
     if(pageViews % 2 == 0)
       //output even number of pageviews
  }
}

Barrakketh

Apothe0sis wrote: »

Familiarising myself with Python and loving it...

Is there a way to call a specific attribute (this is the wrong term, isn't it?) or method from a variable like so:

Like,

>>> b.foo
"bar"

Can you do something like

>>> a = "foo"
>>> b.$a
"bar"

or something so that a is interpreted as an attribute or method name?.

It's not the cleanest looking thing, but:

b = Bar()
b.foo()
foo = getattr(b, "foo")
foo()
getattr(b, "foo")()

If you inspect it you'll see something like:

<bound method Foo.bar of <__main__.Foo instance at 0x03497B98>>

Puddlesworth

Barrakketh wrote:

Apothe0sis wrote: »

Familiarising myself with Python and loving it...

Is there a way to call a specific attribute (this is the wrong term, isn't it?) or method from a variable like so:

Like,

>>> b.foo
"bar"

Can you do something like

>>> a = "foo"
>>> b.$a
"bar"

or something so that a is interpreted as an attribute or method name?.

It's not the cleanest looking thing, but:

b = Bar()
b.foo()
foo = getattr(b, "foo")
foo()
getattr(b, "foo")()

If you inspect it you'll see something like:

<bound method Foo.bar of <__main__.Foo instance at 0x03497B98>>

An if you're stll looking for another way you can even do

>>> b.__dict__["foo"]
"bar"

DrunkMc

Hey guys, I have a java data structure question. I always find it hard to pick the most efficient one for what I'm doing. So I'm doing a flood_fill algorithm. http://en.wikipedia.org/wiki/Flood_fill The bottom one under Most Practical.

I got it working but it was slow as shit. I was using a LinkedList<Pixel> as the queue because I knew I would be removing alot and figured that would be efficient. I was wrong, mainly because I check a lot to see if a pixel is already in the queue. (Because I'm doing an 8-way fill, not 4).

So I implemented Comparable<Pixel> in my Pixel class and switched it to a TreeSet<Pixel> and it sped it up considerably but its still kind of slow. What is a good structure to house Pixels (double x, double y), where you will do many contains and removes and adds? Is TreeSet the best I'm going to do?

Thanks!

Oh, and the images I'm working on are ~3kx3k.

Phyphor

1) You can implement the list yourself and have each pixel object keep track of it's next/previous pointers. Testing for already being in the list becomes constant time.
2) If you can't do that, use an array of bools or something to keep track of the insertion status of your pixels to reduce the time checking if it's already in the list and a stack for your list

Also, you can try to use a further optimization of that algorithm: keep track of runs on the above/below pixels and only add one pixel per run to minimize list management overhead

DrunkMc

Phyphor wrote:

1) You can implement the list yourself and have each pixel object keep track of it's next/previous pointers. Testing for already being in the list becomes constant time.
2) If you can't do that, use an array of bools or something to keep track of the insertion status of your pixels to reduce the time checking if it's already in the list and a stack for your list

Also, you can try to use a further optimization of that algorithm: keep track of runs on the above/below pixels and only add one pixel per run to minimize list management overhead

Really good ideas. With our special case, a co-worker actually thought of something that's similar to your #2. In my case, the images are a byte image, but we're only using 0 or 1. He had the idea of when adding to the queue, change that value to 2, and only go from 1 to 0 when you pop it off the queue. Then instead of structure.contains(pixel), I just have to see, pixels[x][y]!=2. Should be much much faster!

If we didn't have this special case of 0 or 1, I'd def. do #2.

Thanks!

Update:
In case anyone is interested, with using the image as an indicator as to whether a pixel was already on the queue, instead of doing a .contains() on the queue, the flood delete went from ~50% of the processing to 7% of the processing. And thats in the worst case sceneario. So, Woo! Interesting lesson learned!

JediNight

Anyone care to take a swing at pointing me in the right direction for a Java assignment? I've tried fixing my issues for several hours now and I keep hitting new snags related to the problem.

Basically atm it's getting to the last element in the ArrayList and hanging with an ArrayList<E>RangeCheck error. The odd thing about it is that it correctly pulls the customer name from that row of the ArrayList then hangs on pulling the sales data when no new incrementing has occurred.

Also maybe thoughts on my approach atm: I'm parsing an invoice list and putting it into an ArrayList and sorting it. Then going back through it to process the report and generate a total sales report for each warehouse and grand totals. Customer names appear multiple times in the sample sales DAT from the professor, so when its sorted they are clumped together (hence checking to see if the next name is the same).

http://www.privatepaste.com/2ac2c1ac0c

End

JediNight wrote: »

http://www.privatepaste.com/2ac2c1ac0c

I can't stop looking at this:

if (moreCustomers = true) {

edit: but...I can't see where moreCustomers is ever set to anything but true anyway

Incindium

Dude you are accessing the next element of the arraylist even when you are on the last element with your while

while (arrayList.get(count).getCustName().equals(arrayList.get(count+1).getCustName())) {

Not to be mean but that whole thing needs to be reworked a bit. Why don't you use for each on your arraylist
and then reference the single Invoice object rather than looking it up from the arraylist for every method call?

IE

for (Invoice myinv : arrayList )
{
myinv.getWarehouse(); //or whatever
myinv.getCost();
}

etc.

JediNight

End wrote:

JediNight wrote: »

http://www.privatepaste.com/2ac2c1ac0c

I can't stop looking at this:

if (moreCustomers = true) {

edit: but...I can't see where moreCustomers is ever set to anything but true anyway

Thats a code fragment from trying to fix my problems. I'll clean it up later.... Since it doesn't harm anything atm and I wasn't sure if I would need to re-implement it later I left it for now.

Incindium: Thanks for pointing out the issue with the count+1. Should I make some kind of null check then? And I'll read up on foreach as well. This is the start of 2nd semester Java for me, so I don't know all of the ways to properly code yet. (Especially since most of my class is getting remedial review from the Professor while I work ahead and try to learn better ways of accomplishing the Projects. Pisses me off actually, because she's basically reteaching 1st semester to them while I pay for it...)

EDIT: Would I be better off scrapping the entire idea of parsing and sorting the invoice file first? And just going with something like a binary search for a matching customer name when each new line is read from the invoice file and add that data to it?

Monkey Ball Warrior

Foreach is a beautiful thing. It nukes a lot of off-by-one bounds errors that otherwise pop up when you are learning a new language (or to code at all).

As a heads up, it looks like this in the languages I know and that have it:

Java

for(ThingType thing : box)

C#

foreach(var thing in box)

Javascript

for(var thing in box)

Go

for thing, index := range box

The only catch is that you lose the index number (except in Go). Most of the time you don't need it, but occasionally you do and have to go back to the traditional for(int i = 0; i < x; i++) format.

In Java you can do that with arrays and anything that implements Iterable

Phyphor

foreach in pre-C++11!

#define foreach(obj, ...) for(__VA_ARGS__::iterator it = (obj).begin(); it != (obj).end(); ++it)

admanb

Frem

Monkey Ball Warrior wrote:

The only catch is that you lose the index number (except in Go). Most of the time you don't need it, but occasionally you do and have to go back to the traditional for(int i = 0; i < x; i++) format.

It's also easy to get the index number in Python:

for thing, index in enumerate(box):

Monkey Ball Warrior

Frem wrote:

Monkey Ball Warrior wrote:

The only catch is that you lose the index number (except in Go). Most of the time you don't need it, but occasionally you do and have to go back to the traditional for(int i = 0; i < x; i++) format.

It's also easy to get the index number in Python:

for thing, index in enumerate(box):

Yeh I really need to sit down and learn python at some point.

Weretaco

You can retain it in c# enumerables too with little work and anonymous types.

Infidel

Phyphor is good people.

Barrakketh

Frem wrote: »

Monkey Ball Warrior wrote:

The only catch is that you lose the index number (except in Go). Most of the time you don't need it, but occasionally you do and have to go back to the traditional for(int i = 0; i < x; i++) format.

It's also easy to get the index number in Python:

for thing, index in enumerate(box):

Umm...you have it backwards:

for idx, thing in enumerate(box):
    stuff

It also supports a start parameter in case you want to start counting from something other than 0.

templewulf

Ugh. Am I wrong in thinking there is no reason that dictionaries aren't serializable in C#?

I'm trying to save sprite sheet metadata to XML files with XNA. Does anyone have any suggestions for how to do this now that I can't just serializer.Serialize()? I'm not married to XML either, but I'd like to just implement a ready-made solution and get on with the interesting parts.

Infidel

templewulf wrote:

Ugh. Am I wrong in thinking there is no reason that dictionaries aren't serializable in C#?

I'm trying to save sprite sheet metadata to XML files with XNA. Does anyone have any suggestions for how to do this now that I can't just serializer.Serialize()? I'm not married to XML either, but I'd like to just implement a ready-made solution and get on with the interesting parts.

They aren't XML serializable out of the gate, but you can do a few things.

One would be to do a binary serialize of the dictionary object and then stuff that as binary data into a member that then is serialized by your overall XML process.

Another would be to implement your own SerializableDictionary from Dictionary and IXmlSerializable.

Monkey Ball Warrior

Yeh I hit that little snag when trying to save dictionaries in Profiles in ASP.NET. (that and the fun that is specifying generic types in web.config) I ended up just specifying binary serialization and it worked out fine...

...except come to find out I didn't have anywhere to actually put the profile data since we use Oracle instead of MS SQL server at work and the backflips you have to perform to get that working ended up being incompatible with other weird things that we are doing with Oracle.

tl;dr Oracle is weird.

Alistair Hutton

Infidel wrote:

templewulf wrote:

Ugh. Am I wrong in thinking there is no reason that dictionaries aren't serializable in C#?

I'm trying to save sprite sheet metadata to XML files with XNA. Does anyone have any suggestions for how to do this now that I can't just serializer.Serialize()? I'm not married to XML either, but I'd like to just implement a ready-made solution and get on with the interesting parts.

They aren't XML serializable out of the gate, but you can do a few things.

One would be to do a binary serialize of the dictionary object and then stuff that as binary data into a member that then is serialized by your overall XML process.

Another would be to implement your own SerializableDictionary from Dictionary and IXmlSerializable.

Or you can XML serialise a list of DictionaryEntry objects and then reconstruct the dictionary with them on deserialisation.

bowen

Alistair's is probably the quickest to do.

Jimmy King

mmmm, foreach. I just had to go through a bunch of javascript at the new job written by an old C/C++ dev and turn for loops into foreach because there were off by one errors in all of them and they required manual fiddling any time new category thingies get added to the site. He's a smart guy, wrote multithreaded c++ stuff that works afaik, so just not up on this fancy newer stuff.

btw, be careful with the "for(var thing in box)" style of foreach in js. If you're iterating over an Array(), with "for(var thing in box)" you're actually getting the keys because it iterates over the object's dictionary (or something like that). That shit gets me every damned time. JavaScript also has a proper Array.forEach() which returns the values rather than the keys/indexes but IE didn't support it until IE9 so it's not terribly useful.

  myArray = new Array();
  myArray[0] = "zero";
  myArray[1] = "one";
  myArray[2] = 2;
  myArray[3] = 3;
  myArray[4] = 4;

  myArray.forEach(function(x) { alert(x)});

 /* or if you also want the index, similar to python you can do this.  x and idx may be backwards*/
 myArray.forEach(function(x, idx) {alert("Index is " + idx + " and value is " + x);