To be honest that's pretty much what my first 5 minutes in any command line looks like.
Also shitting on a woman who created camps and gives scholarships to teach young women to code is kind of /r/KIA.
I've actually been watching Karlie Kloss's coding adventures with some interest after a former student of mine popped it on my Facebook. I'm still struggling to determine whether she's really programming or whether she's just fronting for some non-profits/companies who are using her as a spokeswoman. Not that the latter's a bad thing; I've just been wondering whether she's actually learning programming or just advocating that others do so.
Nothing I've seen has convinced me that she's actually done any programming, which is disappointing. I've seen some interviews, but it sounds mostly like she's just blathering generic "computers are important!" technospeak. I've seen pictures of her at the bootcamp, but again, that doesn't mean anything. If this picture was from 30+ weeks ago (Am I instagramming right?) and there's a ... box in the top left that says "nicely placed product" there (uhh, is this some sort of ironic thing that kids do nowadays), then I'm skeptical. I mean, if it takes a supermodel frontwoman to convince young girls to go learn computer programming skills, then yay, but I'd be a little grossed out if they're putting all this energy into maintaining this facade that she's actually programming when she's not. Especially since that also plays into the second-level stereotype which is that all women programmers (and engineers) have to be super-professional-incredible-awesome-super-hot-cool.
Edit: Can't find it now, but I once had a link to a really nice piece about how there didn't just need to be more amazing women in the tech industry; there needed to be more average women in tech, that parading out an endless line of super-put-together-over-achieving women in tech won't really fix the underrepresentation problem, since we'll run out of them at some point, and what we really want is for women to go into tech without requiring them to be leagues ahead of the average male worker. We want women in tech to be normal, not exceptional, and they should be allowed to be "normal" rather than living with this implied burden that they must represent their whole gender with the utmost perfection.
The memset-transposed-args warning/error in GCC 5.x is the best/worst thing when building other people's code. At least none of the ones I've found yet are security issues?
a5ehren on
0
Options
GnomeTankWhat the what?Portland, OregonRegistered Userregular
We're starting to evaluate Mesosphere DC/OS and Docket Data Center for container orchestration at work. Any of you have experience with either?
We're starting to evaluate Mesosphere DC/OS and Docket Data Center for container orchestration at work. Any of you have experience with either?
Not corporate so haven't done eval on the enterprise solutions, but we've been running bare docker engine in production for a while now and have moved other environments to swarm clusters. A couple of months ago did in-house eval on kubernetes/mesos(marathon + chronos only) & ecs(good, but only if you are already on aws). Feel free to toss questions, overall we had Mesos as a very clear winner because of ease of setup and API. UI(if of importance) is brutal over the board in the open source versions. In the end, we're still in a decision limbo, simply because at our scale(tens of machines), almost everybody around us is focusing on kubernetes and it feels intimidating.
The DDC didn't even exist then, so no impressions on it. While I'm a big fan of the docker engine, I really think they missed the boat on monetising on that front.
Any suggestions on a regex I can use in the following scenario?
M84569A Pathological fracture in neoplastic disease, unspecified tibia and fibula, initial encounter for fracture
I want to grab the very first space and ONLY that space. That space has one or more spaces until the description.
(\s+)
Gives me every instance. I cannot figure out how to just grab the first instance.
And please. Don't talk about how regex sucks. I know it does. This is just to generate a quick script and not going into any sort of production environment.
Did you try an anchor? /^\S+(\s+)/
That says, "starting at the beginning of the string (or the line), find everything which is not a space, followed by everything which is a space". That will capture only the first contiguous section of whitespace.
I think I'll just wait until she sends me the list using a comma instead of a space to separate the two. This text file is 69824 lines long. And I'm spending 20 minutes trying to generate a regex when I could be working on not-dumb shit.
We're starting to evaluate Mesosphere DC/OS and Docket Data Center for container orchestration at work. Any of you have experience with either?
Not corporate so haven't done eval on the enterprise solutions, but we've been running bare docker engine in production for a while now and have moved other environments to swarm clusters. A couple of months ago did in-house eval on kubernetes/mesos(marathon + chronos only) & ecs(good, but only if you are already on aws). Feel free to toss questions, overall we had Mesos as a very clear winner because of ease of setup and API. UI(if of importance) is brutal over the board in the open source versions. In the end, we're still in a decision limbo, simply because at our scale(tens of machines), almost everybody around us is focusing on kubernetes and it feels intimidating.
The DDC didn't even exist then, so no impressions on it. While I'm a big fan of the docker engine, I really think they missed the boat on monetising on that front.
I think we're pretty sold on buying one of the enterprise solutions at this point (they are actually pretty aggressively priced at this point, not over the top). We're more in the 30-50 node range, so starting to get in to that area where managing it can be rough without a nice orchestration dashboard.
Were you planning to use Mesos to drive your swarm clusters, or were you planning to use Kubernetes as your swarm coordinator instead of docker swarm?
not a doctor, not a lawyer, examples I use may not be fully researched so don't take out of context plz, don't @ me
0
Options
GnomeTankWhat the what?Portland, OregonRegistered Userregular
edited April 2016
You're tired of having to run all your applications on special snow flake servers, with locked library and framework versions, and the lack of developer freedom that offers? Equally your ops people are tired of having to build special snow flake servers, with locked library and framework versions, every time a developer wants to try something new? That's pretty much the point of going to a container based architecture. There are a lot of side benefits, like making CICD a lot easier and helping with data center utilization...but really it's about letting developers worry about how their app runs, while ops worries only about the container hosts.
e: They are also really important if you're going to a highly decoupled system, or a micro-service architecture.
Yeah I control both and it's not really a big deal to me TBH.
The only thing I can think of that'd benefit from it is my RESTful system, but it's integrated with some windows-centric stuff that outside applications use. I could probably decouple it but meh.
Maybe going forward I'll write new RESTful services with docker in mind. Is there a good core .NET base image yet?
not a doctor, not a lawyer, examples I use may not be fully researched so don't take out of context plz, don't @ me
0
Options
GnomeTankWhat the what?Portland, OregonRegistered Userregular
Yes, microsoft/dotnet is the official CoreCLR base image.
Has Docker yet done anything to encourage better security practices? I know it's not their fault that people grab code from public repositories and throw it into production without checking it, but they still seem shy about admitting and warning that the container runs as root, and their goddamn install process still pipes curl to a shell.
0
Options
GnomeTankWhat the what?Portland, OregonRegistered Userregular
edited April 2016
Containers don't need to run as root anymore using user namespaces. On top of that any enterprise that just lets people pull random images from the hub in to their production environment is dumb. You run a Docker trusted registry inside your firewall and you either white list specific base images, or you provide your own approved base images for various scenarios (this is how we do it...you won't see FROM node:5 in our docker files, you'll see FROM approved/node5, which exists in our trusted registry and is rebuilt and patched regularly).
e: Piping curl/wget to a shell is an incredibly common Linux practice. That said, that's not how we install Docker. We have Chef scripts that build our docker hosts, either a node or a manager. That's all handled by ops. All our docker hosts are effectively exactly a like, only minor differences between what a manager and a node do in terms of how swarm is setup and how the consul servers/clients are allocated.
e2: Also, Docker has never been shy about admitting in a default config the containers run as root...but they, like me, realize how overblown the problem really is. To break out of a container is to break kernel namespaces, which are battle tested and proven. Even if you run Docker as a special 'docker' user, guess what permissions that user needs to have? It's not pure root power, but it's close. That user can create c-groups and namespaces, mess with network interfaces and write to some secure areas of the system. This is unavoidable for a containerization system to work. Even Rocket which touts "not running as root" as a feature has to have effectively a super user to actually work correctly.
Jesus Christ why is vertical alignment so fucking complicated? I have a button that I need to have aligned in a div but I cannot get it in the middle.
You also need to consider line height in a lot of cases. I banged out a SASS vert-align function that sets all the container heights, line heights, and vertical-align. I love and hate front end work.
My usual technique is this. (Drop the "left" rules if it doesn't need to be horizontally centered too.) But I know the more modern approach is to use flexbox, and support for that is pretty good now.
We're starting to evaluate Mesosphere DC/OS and Docket Data Center for container orchestration at work. Any of you have experience with either?
Not corporate so haven't done eval on the enterprise solutions, but we've been running bare docker engine in production for a while now and have moved other environments to swarm clusters. A couple of months ago did in-house eval on kubernetes/mesos(marathon + chronos only) & ecs(good, but only if you are already on aws). Feel free to toss questions, overall we had Mesos as a very clear winner because of ease of setup and API. UI(if of importance) is brutal over the board in the open source versions. In the end, we're still in a decision limbo, simply because at our scale(tens of machines), almost everybody around us is focusing on kubernetes and it feels intimidating.
The DDC didn't even exist then, so no impressions on it. While I'm a big fan of the docker engine, I really think they missed the boat on monetising on that front.
I think we're pretty sold on buying one of the enterprise solutions at this point (they are actually pretty aggressively priced at this point, not over the top). We're more in the 30-50 node range, so starting to get in to that area where managing it can be rough without a nice orchestration dashboard.
Were you planning to use Mesos to drive your swarm clusters, or were you planning to use Kubernetes as your swarm coordinator instead of docker swarm?
We're of similar size(assuming you didn't mean customer facing nodes), but some of the physical nodes are managed by Openstack and split additionally. That said, the current docker cluster size in production is quiet small - 4 nodes only, as a lot of the data nodes run(and will probably stay running) on baremetal. The plan is to migrate the small cluster and then start bringing more resources in from there.
The evaluation "won" with Mesos only for nodes, Marathon for services, delegating networking overlays to docker directly and orchestrating via the API's. The decision was to not use k8s at all. I saw several demos at meetups with kube on mesos and it felt like an overkill.
The thing to keep in mind for our exploration is, that we wrongly approached it with a "Kube's gonna be a clear winner, but let's just do that anyway to have our own opinion and then we continue with it" attitude All the feedback we were getting is that if you aren't in the triple+ digits of nodes and a LOT of services you should find kube easy, right? Well, we fucking didn't and it really put a question mark on the whole thing. I'd expect we'd be running k8s in production by now if we had been happy with the results. Right now it feels a bit like we've halted and still waiting to be convinced we should do k8s instead.
On a side note, I recently had the chance to ask several mid-size companies(100-500 employees) running docker in production if anyone is using kube, and to my surprise nobody actually was, All were engine/swarm only.
Would be very interesting to hear from you an overall impression if you end up going to production with one of paid for services.
A shitload of them. If you have a product team willing to assume responsibility over part of ops, most ICT deps will be more than happy as long as it can be done smoothly. Isolating application deps is super nice, monitoring/logging can be unified (resources and on aps level).....many more.
But hey, the biggest one is...have you ever fucking seen Puppet?
Edit: And yeah, we do the same as Gnome actually. We still provision docker nodes via puppet, but we plan to move to Openstack->KVM->Docker soon.
We're starting to evaluate Mesosphere DC/OS and Docket Data Center for container orchestration at work. Any of you have experience with either?
Not corporate so haven't done eval on the enterprise solutions, but we've been running bare docker engine in production for a while now and have moved other environments to swarm clusters. A couple of months ago did in-house eval on kubernetes/mesos(marathon + chronos only) & ecs(good, but only if you are already on aws). Feel free to toss questions, overall we had Mesos as a very clear winner because of ease of setup and API. UI(if of importance) is brutal over the board in the open source versions. In the end, we're still in a decision limbo, simply because at our scale(tens of machines), almost everybody around us is focusing on kubernetes and it feels intimidating.
The DDC didn't even exist then, so no impressions on it. While I'm a big fan of the docker engine, I really think they missed the boat on monetising on that front.
Popular opinion is Kubernetes sweeps the floor at the moment, running on top of a Mesos cluster.
Currently we're (at the 300 --> up node scale) looking at Mesos, though I'm not honestly convinced that decision totally makes sense yet. Mesos's API is actually pretty rough internally from my current playing with implementing a simple framework, and the docker integration if that's all you're running (and let's face it, that's all you're going to run) is *really* rough (i.e. it translates to command line arguments).
0
Options
GnomeTankWhat the what?Portland, OregonRegistered Userregular
Yeah, I'm pushing pretty hard against Mesos at work. My druthers are to just use docker swarm against the Docker Universal Control Plane. It's so easy to get a docker swarm cluster setup, you can do it in a couple of hours and if the hardware is the right place (aka in the right failure zones and all that data center shit), it's basically highly available and ready to rock. Only thing it's missing is the management UI which is where the UCP comes in.
We'll see though, I'm not the only one making the decision Some good insight though guys, thanks for that.
I have a (hopefully) quick Java question for you guys. I'm doing a homework assignment where I have to extend an AVLTree class to do a few extra things. One of those is to clone an AVLTree in O(n) time.
I'm trying to recursively go down the existing tree by calling a separate function, but I'm getting a really weird error that I don't understand. I'm using this StackOverflow suggestion to design my recursive function. I'm passing through two Nodes of type <Position<Entry<K,V>> to my function, but the error I'm getting is:
"The method copy(<Position<Entry<K,V>>, <Position<Entry<K,V>>) is not applicable for the arguments (<Position<Entry<K,V>>, <Position<Entry<K,V>>).
My function header is: private void copy(Position<Entry<K,V>> oldNode, Position<Entry<K,V>> newNode)
Eclipse suggests that I rewrite this to be "private static copy()" but I don't want to return anything and I lose some generics. Does anyone have any suggestions on what I could do, or a topic I should review?
I know recursion isn't technically O(n) time, but the existing iterator function doesn't return the correct data type for this. The iterator function returns the superclass of the type I need, and I can't get it to cast correctly.
Uhhh. So, you have a Node class, and the Nodes hold Position<Entry<K, V>> objects? Or are your nodes the Position class?
Without seeing more of your code, it looks perfectly fine? The error message looks... wrong? Why are there <>s around the Position, when they're not in your function header?
Also, this method is going to be O(n) time, since you're cloning n nodes + 2 edges max for each. Runtime has nothing to do with whether you run it recursively or not.
Yeah, sorry for not posting more code. We're supposed to extend the AVLTree class as part of the net.datastructures package from our textbook and I don't want to spam a bunch of related classes. You're also right, I mistyped the <> above.
Each Node<K,V> is an extension of a BTNode<Entry<K,V>>, which is an extension of a Position<E> object. Nodes in the Tree contains 4 things: 1 Entry<K,V> object and three Position<Entry<K,V>> objects pointing to parent/left/right nodes. Accessing a Node via getParent(), getLeft(), getRight() returns a Position<Entry<K,V>> object.
Ah, and thanks for correcting my understanding about recursion there - I don't have a computer science background, so I'm still learning a lot.
The full functions that I've written are here:
public static <K,V> AVLTree<K,V> clone(AVLTree<K,V> tree) {
AVLTree<K,V> tmpTree = new AVLTree<K,V>();
// Create empty root node and add to AVLTree
AVLNode<K,V> newRoot = new AVLNode<K,V>();
tmpTree.root = newRoot;
// Recursively loop through 'tree' and add nodes to 'tmpTree'
copy(tree.root, tmpTree.root);
return tmpTree;
}
private void copy(Position<Entry<K,V>> oldNode, Position<Entry<K,V>> newNode) {
// Loop down left-hand side of tree
if(oldNode.getLeft() != null) {
// Create empty AVLNode to store information and add
AVLNode<K,V> tmpLeft = new AVLNode<K,V>();
newNode.setLeft(tmpLeft);
tmpLeft.setElement(oldNode.element());
tmpLeft.setParent(newNode);
copy(oldNode.getLeft(), tmpLeft.getLeft());
}
// Loop down right-hand side of tree
if(oldNode.getRight() != null) {
// Create empty AVLNode to store information and add
AVLNode<K,V> tmpRight = new AVLNode<K,V>();
newNode.setRight(tmpRight);
tmpRight.setElement(oldNode.element());
tmpRight.setParent(newNode);
copy(oldNode.getRight(), tmpRight.getRight());
}
}
Okay, I'm going to assume that you're coping the nodes' contents properly. I'm guessing that's what setElement does, but I don't really know what an element is, so whatevs.
The problem I see is that you're just calling copy like it was a global function. There are no global functions in Java; everything is a method that is either attached to an object, or it's a static method attached to a class. In this case, since clone is already static, you might as well also make copy static. Then you should call copy using AVLNode.copy(...) - replacing AVLNode with whatever class this method is in.
I'm also assuming I'm copying the contents correctly, at least until I get a print() function working.
Ah, interesting... I thought I could just directly call any function within the same class. In that case, if I change the function header to:
private static <K,V> AVLNode<K,V> recurseCopy(BTPosition<Entry<K,V>> oldNode, BTPosition<Entry<K,V>> newNode)
And simply 'return null', the error goes away. I'll have to debug to make sure everything else works, though. Thanks for the help!
Yep, you can't call that copy function as it is written unless you construct a new object of the class it is contained within, which I assume is a Node of some kind.
Rewriting to static should work, and if you don't want to return anything, you can make the method static void.
Posts
I've actually been watching Karlie Kloss's coding adventures with some interest after a former student of mine popped it on my Facebook. I'm still struggling to determine whether she's really programming or whether she's just fronting for some non-profits/companies who are using her as a spokeswoman. Not that the latter's a bad thing; I've just been wondering whether she's actually learning programming or just advocating that others do so.
Nothing I've seen has convinced me that she's actually done any programming, which is disappointing. I've seen some interviews, but it sounds mostly like she's just blathering generic "computers are important!" technospeak. I've seen pictures of her at the bootcamp, but again, that doesn't mean anything. If this picture was from 30+ weeks ago (Am I instagramming right?) and there's a ... box in the top left that says "nicely placed product" there (uhh, is this some sort of ironic thing that kids do nowadays), then I'm skeptical. I mean, if it takes a supermodel frontwoman to convince young girls to go learn computer programming skills, then yay, but I'd be a little grossed out if they're putting all this energy into maintaining this facade that she's actually programming when she's not. Especially since that also plays into the second-level stereotype which is that all women programmers (and engineers) have to be super-professional-incredible-awesome-super-hot-cool.
Edit: Can't find it now, but I once had a link to a really nice piece about how there didn't just need to be more amazing women in the tech industry; there needed to be more average women in tech, that parading out an endless line of super-put-together-over-achieving women in tech won't really fix the underrepresentation problem, since we'll run out of them at some point, and what we really want is for women to go into tech without requiring them to be leagues ahead of the average male worker. We want women in tech to be normal, not exceptional, and they should be allowed to be "normal" rather than living with this implied burden that they must represent their whole gender with the utmost perfection.
I am certain there is a line you could add to you .bshrc to have the prompt display your pwd
...which I realize is moot. I do git branch/status/log one after the other constantly
https://docs.google.com/spreadsheets/d/1OLcAGbXhWIVcl5IziVpG0eKFJS3xi_Sac9kYMkRFvD8/edit?usp=sharing
Not corporate so haven't done eval on the enterprise solutions, but we've been running bare docker engine in production for a while now and have moved other environments to swarm clusters. A couple of months ago did in-house eval on kubernetes/mesos(marathon + chronos only) & ecs(good, but only if you are already on aws). Feel free to toss questions, overall we had Mesos as a very clear winner because of ease of setup and API. UI(if of importance) is brutal over the board in the open source versions. In the end, we're still in a decision limbo, simply because at our scale(tens of machines), almost everybody around us is focusing on kubernetes and it feels intimidating.
The DDC didn't even exist then, so no impressions on it. While I'm a big fan of the docker engine, I really think they missed the boat on monetising on that front.
My prompt is:
(needs git-aware)
Edit:
I want to grab the very first space and ONLY that space. That space has one or more spaces until the description.
Gives me every instance. I cannot figure out how to just grab the first instance.
And please. Don't talk about how regex sucks. I know it does. This is just to generate a quick script and not going into any sort of production environment.
(So that the greedy .* match will eat all the other spaces?)
Did you try an anchor? /^\S+(\s+)/
That says, "starting at the beginning of the string (or the line), find everything which is not a space, followed by everything which is a space". That will capture only the first contiguous section of whitespace.
I've got one right here for you:
" "
I think we're pretty sold on buying one of the enterprise solutions at this point (they are actually pretty aggressively priced at this point, not over the top). We're more in the 30-50 node range, so starting to get in to that area where managing it can be rough without a nice orchestration dashboard.
Were you planning to use Mesos to drive your swarm clusters, or were you planning to use Kubernetes as your swarm coordinator instead of docker swarm?
e: They are also really important if you're going to a highly decoupled system, or a micro-service architecture.
The only thing I can think of that'd benefit from it is my RESTful system, but it's integrated with some windows-centric stuff that outside applications use. I could probably decouple it but meh.
Maybe going forward I'll write new RESTful services with docker in mind. Is there a good core .NET base image yet?
e: Piping curl/wget to a shell is an incredibly common Linux practice. That said, that's not how we install Docker. We have Chef scripts that build our docker hosts, either a node or a manager. That's all handled by ops. All our docker hosts are effectively exactly a like, only minor differences between what a manager and a node do in terms of how swarm is setup and how the consul servers/clients are allocated.
e2: Also, Docker has never been shy about admitting in a default config the containers run as root...but they, like me, realize how overblown the problem really is. To break out of a container is to break kernel namespaces, which are battle tested and proven. Even if you run Docker as a special 'docker' user, guess what permissions that user needs to have? It's not pure root power, but it's close. That user can create c-groups and namespaces, mess with network interfaces and write to some secure areas of the system. This is unavoidable for a containerization system to work. Even Rocket which touts "not running as root" as a feature has to have effectively a super user to actually work correctly.
http://phrogz.net/css/vertical-align/
You also need to consider line height in a lot of cases. I banged out a SASS vert-align function that sets all the container heights, line heights, and vertical-align. I love and hate front end work.
We're of similar size(assuming you didn't mean customer facing nodes), but some of the physical nodes are managed by Openstack and split additionally. That said, the current docker cluster size in production is quiet small - 4 nodes only, as a lot of the data nodes run(and will probably stay running) on baremetal. The plan is to migrate the small cluster and then start bringing more resources in from there.
The evaluation "won" with Mesos only for nodes, Marathon for services, delegating networking overlays to docker directly and orchestrating via the API's. The decision was to not use k8s at all. I saw several demos at meetups with kube on mesos and it felt like an overkill.
The thing to keep in mind for our exploration is, that we wrongly approached it with a "Kube's gonna be a clear winner, but let's just do that anyway to have our own opinion and then we continue with it" attitude All the feedback we were getting is that if you aren't in the triple+ digits of nodes and a LOT of services you should find kube easy, right? Well, we fucking didn't and it really put a question mark on the whole thing. I'd expect we'd be running k8s in production by now if we had been happy with the results. Right now it feels a bit like we've halted and still waiting to be convinced we should do k8s instead.
On a side note, I recently had the chance to ask several mid-size companies(100-500 employees) running docker in production if anyone is using kube, and to my surprise nobody actually was, All were engine/swarm only.
Would be very interesting to hear from you an overall impression if you end up going to production with one of paid for services.
A shitload of them. If you have a product team willing to assume responsibility over part of ops, most ICT deps will be more than happy as long as it can be done smoothly. Isolating application deps is super nice, monitoring/logging can be unified (resources and on aps level).....many more.
But hey, the biggest one is...have you ever fucking seen Puppet?
Edit: And yeah, we do the same as Gnome actually. We still provision docker nodes via puppet, but we plan to move to Openstack->KVM->Docker soon.
Popular opinion is Kubernetes sweeps the floor at the moment, running on top of a Mesos cluster.
Currently we're (at the 300 --> up node scale) looking at Mesos, though I'm not honestly convinced that decision totally makes sense yet. Mesos's API is actually pretty rough internally from my current playing with implementing a simple framework, and the docker integration if that's all you're running (and let's face it, that's all you're going to run) is *really* rough (i.e. it translates to command line arguments).
We'll see though, I'm not the only one making the decision
I'm trying to recursively go down the existing tree by calling a separate function, but I'm getting a really weird error that I don't understand. I'm using this StackOverflow suggestion to design my recursive function. I'm passing through two Nodes of type <Position<Entry<K,V>> to my function, but the error I'm getting is:
"The method copy(<Position<Entry<K,V>>, <Position<Entry<K,V>>) is not applicable for the arguments (<Position<Entry<K,V>>, <Position<Entry<K,V>>).
My function header is: private void copy(Position<Entry<K,V>> oldNode, Position<Entry<K,V>> newNode)
Eclipse suggests that I rewrite this to be "private static copy()" but I don't want to return anything and I lose some generics. Does anyone have any suggestions on what I could do, or a topic I should review?
I know recursion isn't technically O(n) time, but the existing iterator function doesn't return the correct data type for this. The iterator function returns the superclass of the type I need, and I can't get it to cast correctly.
Without seeing more of your code, it looks perfectly fine? The error message looks... wrong? Why are there <>s around the Position, when they're not in your function header?
Also, this method is going to be O(n) time, since you're cloning n nodes + 2 edges max for each. Runtime has nothing to do with whether you run it recursively or not.
Each Node<K,V> is an extension of a BTNode<Entry<K,V>>, which is an extension of a Position<E> object. Nodes in the Tree contains 4 things: 1 Entry<K,V> object and three Position<Entry<K,V>> objects pointing to parent/left/right nodes. Accessing a Node via getParent(), getLeft(), getRight() returns a Position<Entry<K,V>> object.
Ah, and thanks for correcting my understanding about recursion there - I don't have a computer science background, so I'm still learning a lot.
The full functions that I've written are here:
public static <K,V> AVLTree<K,V> clone(AVLTree<K,V> tree) { AVLTree<K,V> tmpTree = new AVLTree<K,V>(); // Create empty root node and add to AVLTree AVLNode<K,V> newRoot = new AVLNode<K,V>(); tmpTree.root = newRoot; // Recursively loop through 'tree' and add nodes to 'tmpTree' copy(tree.root, tmpTree.root); return tmpTree; } private void copy(Position<Entry<K,V>> oldNode, Position<Entry<K,V>> newNode) { // Loop down left-hand side of tree if(oldNode.getLeft() != null) { // Create empty AVLNode to store information and add AVLNode<K,V> tmpLeft = new AVLNode<K,V>(); newNode.setLeft(tmpLeft); tmpLeft.setElement(oldNode.element()); tmpLeft.setParent(newNode); copy(oldNode.getLeft(), tmpLeft.getLeft()); } // Loop down right-hand side of tree if(oldNode.getRight() != null) { // Create empty AVLNode to store information and add AVLNode<K,V> tmpRight = new AVLNode<K,V>(); newNode.setRight(tmpRight); tmpRight.setElement(oldNode.element()); tmpRight.setParent(newNode); copy(oldNode.getRight(), tmpRight.getRight()); } }The problem I see is that you're just calling copy like it was a global function. There are no global functions in Java; everything is a method that is either attached to an object, or it's a static method attached to a class. In this case, since clone is already static, you might as well also make copy static. Then you should call copy using AVLNode.copy(...) - replacing AVLNode with whatever class this method is in.
Ah, interesting... I thought I could just directly call any function within the same class. In that case, if I change the function header to:
private static <K,V> AVLNode<K,V> recurseCopy(BTPosition<Entry<K,V>> oldNode, BTPosition<Entry<K,V>> newNode)
And simply 'return null', the error goes away. I'll have to debug to make sure everything else works, though. Thanks for the help!
Rewriting to static should work, and if you don't want to return anything, you can make the method static void.