Blog-posts are sorted by the tags you see below. You can filter the listing by checking/unchecking individual tags. Doubleclick or Shift-click a tag to see only its entries. For more informations see: About the Blog.
As you may have noticed, we are back in our every 2 month release cycle and a new beta is up on the horizon.
As we have noticed, not many of you use alpha builds to test it against your latest an greatest projects. So here is a particular fine alpha version that is our release candidate for beta35.5 scheduled for Monday.
Please give it a test run with a few patches and send us reports on any bug or problem you encounter. Testing is also the perfect excuse to miss any Easter obligation.
Some have reported that they are seeing ~temp files being written on save. We could not reproduce the error here, but we have now an error pop-up to inform you when something goes wrong and the exception that caused the problem will be copied into the clipboard. Open the projects that have that issue and paste the exception message into a new forum thread to help us tracking it down.
Here is something really great. The new Reactive category gives you tools to handle asynchronous events, background calculations and even enables you to build your own mainloop that runs on a different CPU core. But let's start with a pragmatic explanation of what it is:
In a way, this isn't anything new. Event buses or your typical click events are really an asynchronous event stream on which you can observe and do some side effects. Reactive is that idea on steroids. You are able to create data streams of anything, not just from click and hover events. Streams are cheap and ubiquitous, anything can be a stream: variables, user inputs, properties, caches, data structures, etc. For example, imagine your Twitter feed would be a data stream in the same fashion that click events are. You can listen to that stream and react accordingly.
On top of that, you are given an amazing toolbox of functions to combine, create and filter any of those streams.
Since a while VVVV and VL use these so called Observables to handle external events (i.e. mouse, keyboard etc.) and asynchronous data. This was mostly under the hood and the actual operations for observables are hidden in the VL.DevLib. The reason is that out of the box the operations do not go well together with the frame based Update concept of VL because they are intended to be called only once or when something has changed. But as of now we have wrapper nodes for the most common observable operations that do exactly that, listen for change and only rebuild the observables when necessary.
The go to node for handling events is definitely ForEach Region (was Region (Stateful) in earlier versions) in the category Reactive. This region allows you to place any node inside and can also remember any data between two events. There is also one with version Keep that can filter out events using a boolean output. This region is very similar to the ForEach region for spreads, only that its input and output is event values in time instead of slices of a spread.
You can switch or merge event sources:
There are also filtering options with OfType or Where:
Other nodes include Skip, Delay, Delay (Selector), Scan, Switch, ...
If you want to leave the observable world and pass event values to the mainloop use one of the 3 nodes HoldLatest, Sampler or S+H which all behave a little bit different. Depends on what you need:
It's also pretty easy to generate event sources of your own:
As a general advice, only send values of type Record as event data because they are thread safe. If you send values of any Class type be sure that you know exactly what you are doing.
Yep, totally possible and has useful applications. But i am just gonna let this idea sink in for now...
The above just scratches the surface of whats possible with the reactive framework. If you want to know more browse some of the following links:
The pragmatic Rx expert from the quote above:
2 minute introduction to Rx
Visual explanation of the observable operations:
Operator Reference with marble diagrams
Videos from the creator team. Note that IEnumerable is called Sequence in VL and Spread is also a Sequence:
Erik Meijer: Rx in 15 Minutes
Erik Meijer and Wes Dyer - Reactive Framework (Rx) Under the Hood 1
Erik Meijer and Wes Dyer - Reactive Framework (Rx) Under the Hood 2
Introduction to Rx
Midi was released in 1982 and is one of the most successful hardware communication protocols in the world. The simple nature of the protocol makes it easy to implement and even more important, easy to understand for humans.
This makes it a perfect example for the first event based library in VL using the MIDI-Toolkit developed by Leslie Sanford.
Instead of having all settings on one node, functionality is now separate to allow arbitrary combinations.
Device nodes have an enum input for the input/output device driver you want to use. You can have many of them, even for the same driver. Under the hood they will share the actual device driver resource. The driver is opened only if it is necessary, for example if there is an event sink listening to it.
The dynamic device enum will update as soon as a midi device is connected or disconnected to the machine. So no restart required on configuration change:
MidiIn has one observable output for all midi messages received on the given device. MidiOut has one input that accepts an observable to send midi messages to the given device.
Following the midi message structure, there are filters that allow you to select only the messages you are interested in. For example only midi clock messages, or messages on a specific midi channel:
For all midi message types there are specific nodes to read the message content or construct new messages. These are mostly the native methods of the MidiToolkit library.
You can process a midi message (in fact any event) directly as it occurs. The new ForEach region in the Reactive category executes it's patch for each event that is passed in and can transform the event into a different message type and decide whether to pass the current event on via the Keep output.
This is part of a bigger programming paradigm that was also polished for the new midi nodes. Definitely check out for the blog post on Reactive Programming.
At some point all async input event handling in the background will be over and you want to leave the observable world and have the processed values in the main loop. For that there are several options:
For supereasy controller value input there is ControllerState or NoteState:
For more advanced scenarios refer to the Reactive nodes HoldLatest, S+H or Sampler which provide ways to pass event values safely to the mainloop.
If you want to generate midi messages in the mainloop you also have a simple node that generates controller message events:
For other messages use the Reactive nodes ToObservable which create an event source that you can use to send events from the mainloop.
Since VL makes a difference between a single value and a spread of values, some nodes come in 'plural' version to allow listening for example for multiple channels at one.
one of the more basic things any programming library has to support is parsing and the creation of XML data-structures. since vl is based on .net we don't have to invent anything here but can make direct use of .nets XDocuments, XElements, XAttributes datatypes. so we're happy to announce that in cooperation with dominikKoller we added xml/json support for vl:
so basically anything you could already do in vvvv plus some more. and this is only what we brought to the surface for you. using the underlying .net datatypes (XDocuments, ...) directly, a pro-user will (later) easily be able to use the whole range of functions that those datatypes provide for more advanced use-cases.
so, once again, something for the whole family..available in latest alphas now.
Pointmovers of vvvv!
The Editors got some new features.
Stepping through the points via Keyboard
Marquee selection in a Bezier Editor
Support for DX11
The Editors by itself are graphics-engine-independent, but their points, lines and handles must be rendered. From now on there are DX11 Views for every Editor.
Available in latest Alpha builds.
previously on devvvvlopment: vl-networking-and-async
With beta35 (including VL) out in the wild and people apparently starting to use it (see forum threads about it) we're quite happy with the feedback so far. We know of people already including VL in their daily patching and getting the hang of it. Others are still more cautious and are waiting for better documentation or are hoping for node17 to open their eyes. All fair enough, no pressure.. in any case the question is of course:
One of the many aspects that vl will be different when compared to vvvv is its node-library. In vvvv we have a closed-source base-library (written in delphi and closely bound to the vvvv core) plus a wide range of open-source nodes including community contributions (written in c# via a plugin-interface). For vl we'll still distinguish between a base-library and the rest, but all libraries will work the same way, ie not be bound to the core. And not even need a plugininterface. And be open-source.
So from the beginning in vl we separated core from libraries and already have a git-repository including all of VLs libraries organized in neat packages (as you can basically already see in your beta35\lib\packs) that we're ready to open. Why haven't we done so already? Well, by releasing library source-code we're kind of committing to a style that everyone should be able to use to write nodes for vl. Therefore this is really a crucial part that we simply wanted to give a second look.
Remember when at node15 we teased how to define nodes for vl? Everything already felt fairly simple indeed. The typeimporter, a breeze. As mentioned in previous blog-posts we've been continuously importing libraries ourselves since then and noted a few things that we needed to improve to make the workflow for future library developers even more convenient.
So this is what we're reworking at the moment and what we hope to be releasing soon:
This will reduce the barrier for developers enormously because everything they have to do to contribute to the vl nodelibrary, will be very little vl-specific and very close to what any c#/.net developer is doing anyway.
Other than that our focus until node17 will continue to be the integration of vl with vvvv, improving documentation and adding the one or other smaller feature. So, that at node17 we have a strong foundation to teach on and hopefully even already some new, contributed libraries..
Until the next update,
previously on VL: vl-progress-report-4
the vivid blog reader already knows the drill: everything stays the same if you liked it just the way it was.
Specify remote host (IP address), a nice port number, connect some data and bang the send to let your UDP packets travel over the network. Or open a server to receive bytes arriving on the specified port. The only difference to vvvv you might see is, that here you also get infos about the sender of the packets via the Remote Endpoint output (which is an IP Address and a port)
same same for the TCP nodes: The client will try to connect to a server. And once the connection is established, you can send and receive bytes.
The TCP Server awaits incoming connections to talk to. The subtle difference here is the Tuple input, where you would expect the data pin. No one ever requested it, but now you can decide which packet should be sent to which client by specifying IP address and port together with the message. In case you still want to send the same packet to all of your clients, just set the address to 0.0.0.0 and port 0
so why did it take so long, what's the goodies behind that?
Unlike the monolithic networking nodes in vvvv you can peek inside the VL ones. The goal was modularizing on a much lower level to be able to provide the very basics as nodes for the patcher:
Good news everyone, from now on you're getting a brand new node able to talk to your loaded-with-firmata Microcontroller Boards (like Arduino, Teensy, Particle.io, ...):
Hardware for the development was kindly provided by QUADRATURE.
Oh, by the way, these new Firmata nodes are fully implemented in VL.
And of course there are some caveats.
The nodes are now available in Alpha Builds.
This is a rework of the original release of 'Arduino Second Service'.
See the previous (now outdated) blogpost.
this little feature upgrade for both vvvv and vl has been lingering around the alphas for a year already...
do you (have to) create touch applications with vvvv? then read ahead:
Pan, Zoom, Rotate, TwoFingerTap and PressAndTap touch gestures are now handed over to you via the GestureDevice. As expected they're multitouch capable, so they report an ID, the position, the distance (Pan & Zoom amount) or the Rotation, and a device ID (in case you have multiple monitors hooked up)
As with the Touch nodes, these also rely on the windows message loop. so touchscreens with windows capable drivers required. fortunately most screens do have these nowadays...
check the helppatches for a microdemo
hope you like it,
previously on VL: VL: Many Paths Lead to Rome
so while the features are freezing we're still cooking the bytes on a low temperature to get them all smooth..
Finally we took the time to do a few performance tweaks on the UI. Not perfect but quite an improvement to what you've seen so far. Selecting nodes and moving them around is now definitely at interactive rates. What you'll still see is that your animations are interrupted while handling the patch. While this might be annoying in the first version of the VL integration, we have worked on the foundations for a non-blocking approach. So expect that to improve in the future. Just not now, priorities are to make everything work first.
Previously all VL nodes which do time based calculations had to have an input pin for the time per operation. The default value for the Time type was set per frame by the VL runtime.
This is now different: You can make a Clock or a FrameClock input on the Create operation of your patch or type and store the clock in a pad to access the time in the other operations. This prevents you from making time inputs for every operation. The default value of the Clock if nothing is connected is the VL runtime frame clock.
The Clock type is very simple and only has a Time property.
The FrameClock has a additional property TimeDifference which holds the delta time since the last frame. This saves you from calculating the delta time everywhere on your own.
If all goes well we'll see the beta35 (including VL as a first class patching language) release soon™ and with it we're entering phase 2 where we'll actively try to convince you to have a look at VL and actually start using it for your projects. At the risk of repeating myself.. this should be done by updating the gray book and providing a couple of introductory video tutorials. And then the world...
If you want to get started with VL right now, here is what you do:
If you want to let us know that you like anything you just read...we're always up for a comment or even a flattr:
anonymous user login