Hacker Newsnew | comments | show | ask | jobs | submit | flixic's comments login

For anyone late, I have 9 invites. My details are on keybase: https://keybase.io/lekevicius

-----


And if there's anyone even later, I've got a few as well: https://keybase.io/oddevan

-----


Be sure to read how it works: https://plus.google.com/+NiklasvonHertzen/posts/Qo6T85W8sFQ

Very smart and creative solution.

-----


As a citizen of Lithuania, I disagree. You can feel destabilization efforts here.

-----


Thanks. I am aware of the issue, and wasn't able to fix it easily, but will try. Also I didn't spend much time with web version, because on mobile there are much better native apps. Actually, I might even remove the Play Online option from mobile.

-----


Hi! Thanks for feedback.

- I deliberately de-emphasize the Play Online option. This game is much, much more fun to play on a smartphone. I even thought about not having web version at all. If you have a smartphone, try downloading the game.

- Generally people are happy about "no instructions at all" design, and that again is deliberate. I'll think about improving tutorial ramp a bit, there are some things to improve.

- This design choice comes from mobile. There I want to show the completed design so you could screenshot it and share it, or just use as wallpaper.

-----


I like no instructions design as well. What I meant was change the first level so you can win it within one or two clicks without even trying to make it more obvious what the goal is.

-----


There is no score, and because levels are random, number of moves is not important. I plan to add level numbers in version 2.

-----


This game is addictive. I played it and loved it. I like to play simple games.

But I feel this do needs some kind of gamification and proper ordering of levels. As twice I received simple puzzle after completing some harder!

Plus if possible do change some colors once the loop is completed. I felt those colors are like olden days. Rest whole game has cool colors.

-----


I like it just as it is, without score and with random levels. I think that having an easier level after a harder one makes me want the play it even more. Please author, if you add those suggestions, maybe add them as a different play mode, since some of us enjoy it exactly as it is. So much so, that I've rated the game on Google play, I think it's my first one.

-----


Even adding another play mode would be too much. I want this to be the most minimal game it can possibly be.

There were quite a few suggestions for level number, so you could feel your progress, so that might be the only visible feature in version 2.

-----


Does this mean the levels are randomly selected, or are they randomly-generated?

-----


Almost all are randomly generated. First 11 tutorial levels are predefined, as well as a few other predefined levels sprinkled over the next couple hundred levels.

-----


Avid chess player and a kickboxer? He could be a really good chess boxer: http://en.wikipedia.org/wiki/Chess_boxing

-----


I was hoping this would be a team sport:

- Each team consists of a chess player and up to 16 boxers

- When one piece captures another, their corresponding boxers fight each other

- The winner of the bout determines which piece is removed from the chess board

-----


That's an interesting idea. The main problem I see is that it would take away a lot of the advantage of taking pieces.

-----


Good point. You'd probably want to match "piece strength" with weight class then, so that a pawn vs. queen is not a fair fight by any means.

-----


That would go some way towards solving it. But what about e.g. a pawn taking a pawn? It's an interesting thought experiment...

-----


But 8 of the boxers would be pawns, so how do you make sure the attacking pawn wins at least half the time?

-----


This is essentially the idea the computer game Archon was based on.

-----


After we were acquired, I designed an extensive timeline of our company history, people and apps released. It sometimes looks bad on mobile, but in general works pretty well, loading data from JSON.

http://lemonlabs.co/timeline.html

-----


This looks great, have you ever thought about open sourcing it?

-----


CSS and JS layout is a mess full of hacks, but I thought about cleaning up and open-sourcing for more general uses. One day (:

-----


Please, don't go that route. Rather, make this reusable for others right now (by making it Free Software), and accept help from outside for the cleanup.

To put this in Eric S. Raymond's words:

Release early, release often!

-----


I hereby declare all original code on http://lemonlabs.co/timeline.html as licenced under MIT, (c) Jonas Lekevicius.

...it's just that code there is compiled from CoffeeScript, concatenated, uglified and so on. Others will have very hard time exploring it.

-----


Thanks Jonas. Unfortunately I think vog is right. It was be awesome if you FOSSed the original source.

-----


I doubt this is of any help for anyone, without having access to the actual source code (and having that licensed as FOSS).

Nevertheless, thanks for trying.

-----


You are completely right. But doing more requires more time, and that I only can promise One Day™.

-----


From the folder icons in resource zips you can see that they are using Google Drive.

Interesting, considering the size of the corporation.

-----


What's the issue with Google Drive? Do you suggest they should come up with their own whole backup solution? Why reinvent the wheel anyways. Not to mention it's possible that only certain teams use it, not all the teams, nor the teams who have truly sensitive information in the spirit they don't want Google to see.

-----


No issue, just interesting.

I would assume that such a big corporation would have very strict data sharing policies. Apparently, that's not the case. Interesting!

-----


I couldn't find the folder icons. Could you post the link to it please?

-----


Well this is interesting.

Their FAQ[0] contains some more information, but still does not explain everything. They say "however, instead of being installed on your local machine, it is running in a virtualized environment so can be accessed from any Chrome browser or Chromebook. Because this version of Photoshop is running in a virtualized environment, you open, save, export and recover files from/to your Google Drive rather than your local file share."

It would seem that they are indeed streaming the video (VNC-style). More over, space requirement is only 350MB (Photoshop is normally much bigger), and there's also this:

"If network connectivity is lost, you will need to launch a new session. A recovery folder called ‘Photoshop Recovery’ will be created in the root of your Google drive. To recover files, simply double click to open a file."

Overall, this doesn't sound good at all. If they are streaming screen "video", color correction and pixel-level precision in design is going to be tough. Photoshop seems like one of the most difficult programs to work via VNC.

[0]: http://edex.adobe.com/projectphotoshopstreaming/faq

-----


Whose to say this is VNC-based? VNC is very old technology. RDP/NX solve a lot of these problems. Or this could a x-server like application. Who knows, but I doubt Adobe is launching something that uses VNC-type technology where it samples and over-compresses the screen. I imagine things like color precision are well taken care of.

Also, anyone else notice this amusing note at the bottom of the FAQ:

Please post your issues to [Insert Forum Link]

-----


If you were building this for a modern Linux client, wouldn't spice be the logical technology to leverage?

https://en.wikipedia.org/wiki/SPICE_%28protocol%29

-----


Or the server/client both implement custom compression on top of VNC.

-----


The color correction can be done on the server. It just needs to get and upload your monitor profile, and Photoshop on the server returns color corrected pixels. Of course, Chrome cannot even read the monitor profile on Linux yet. I'm not sure if this also applies to Chromebooks though. Since it's a fixed device, it probably ships with the monitor profile file factory stored somewhere. Or Adobe gets the files for the few dozen or so Chromebooks from Google, and the browser only sends an identifier.

-----


And then H.264 artifacts probably undo that color correction.

-----


I guess they are using VP9, which has 4:4:4 chroma subsampling: http://youtu.be/xo_R40C7RTo?t=3m35s

-----


Correction: H.264 also has 4:4:4. If you see artifacts, that probably means it wasn't correctly compressed.

-----


Adobe has spent decades worrying about color accuracy in their applications in every path it can take to your eyeball and they're not just going to screw up the last hop because it didn't occur to them to take the effect of whatever remoting technology they're using into account.

-----


Based on the backlash over Creative Cloud, I don't think they really care.

-----


Has there been backlash over color accuracy in Creative Cloud?

Because there are a lot of things Adobe might not care about as much as you'd like- but color accuracy is one of the things they care about A LOT.

(I spent a summer working in an art computing lab, nearly every minute of which was spent managing color accuracy)

-----


  Photoshop seems like one of the most difficult programs to work via VNC.
I think it's actually the opposite. Usually only a small amount of the screen is changing, so response is fast. For full-screen or complicated effects they can provide more CPU horsepower than your average laptop, so the operation may actually complete faster, even if screen redraw takes more time.

FWIW I've used Photoshop over Remote Desktop in the past and with a good connection it feels perfectly usable. There's a little input lag which is only noticeable with sweeping brush effects, not UI interactions, and like I say large area screen redraw (e.g. scrolling) suffers. There's no visible compression or artifacts and I don't see why color correction would be an issue.

-----


> Overall, this doesn't sound good at all. If they are streaming screen "video", color correction and pixel-level precision in design is going to be tough. Photoshop seems like one of the most difficult programs to work via VNC.

This would be a huge issue for the professional market. But it appears that they are targeting education with this first (or maybe just that market). Color precision is likely not the highest of priorities.

-----


I suspect that it's built using Native Client (https://developer.chrome.com/native-client), not using VNC. "Streaming" in this context probably means that the code and files are dynamically served by the server, rather than referring to streaming video.

-----


No, it's actually streaming. As the link says, they run managed vm's and stream it. As someone else says, VNC is really old technology. We can do better.

-----


Nothing in this discussion indicates that anyone knows what streaming technology they're using. However of course there are better and worse technology. Streaming RDP from a Windows box a thousand miles away, for instance, is a world more enjoyable and usable than VNCing to the Mac box connected via a 1Gbps twisted pair. App level virtualization through the same, interacting and engaging with the rest of your desktop, is close to magical.

-----


"Nothing in this discussion indicates that anyone knows what streaming technology they're using"

I actually do, but i don't know whether it's public, so i don't talk about it.

However, i'll point out that Google already has something called Chrome Remote Desktop that already does streaming desktops.

-----


IS it possible that there is a localized virtual machine setup and the image itself is what is streamed over the network? IE you would have a locally installed version of linux on a virtual machine and an additional drive is essentially steamed over the network via SMB os something that contains the PS application?

-----


I can't see any practical reason why this would be the case. Photoshop is a pretty hefty application in terms of RAM/CPU, so it makes a lot more sense to run it remotely, since Chrome Books are built to be cheap, not powerful.

-----


I agree, although perhaps ChromeBooks are "cheap and weak" because there are no applications that use the power (hence the failure of the Chrome Pixel). If ChromeBooks are to become ubiquitous, I think local VMs is a great solution for legacy applications.

-----


Or perhaps NaCl?

-----


I wonder if they are going to pay for all the windows or mac licenses to run on or they have a linux version working inhouse.

-----


Also, I wonder how this affects bandwidth. Are you just constantly streaming HD? Are there latency issues? What about if the network gets congested? It sounds pretty terrible.

Also on the bottom of that FAQ it says "Please post your issues to [Insert Forum Link]" heh.

-----


"Also, I wonder how this affects bandwidth. Are you just constantly streaming HD? Are there latency issues? What about if the network gets congested? It sounds pretty terrible."

I can imagine, given experience with VNC or even some crappy RDP implementations, it sounds like it would be terrible. But those are pretty old and bad technologies for remote desktop usage, with known issues. You can do better.

I'm sure there are situations where the experience is not ideal.

But remember that you are talking about people using chromebooks here. They are computers basically built with the expectation that you have good internet, and they are targeting the education market here.

I would point out that neither Adobe, nor Google, would want people in the education market (one of Adobe's prime targets) to have a crappy experience with Photoshop, as it would be really bad for both parties (if Google makes a bad first impression with chromebooks, people write them off and never try them again. If Adobe makes a bad impression with photoshop, people won't pay hundreds later on to buy it).

So one would imagine they would not release a terrible experience :)

-----


the expectation that you have good internet, and they are targeting the education market

Sadly I don't think I've ever been in an educational institution where the wifi was particularly reliable. Wired ethernet yes, so a computer-lab setup can work well with remote-desktop type stuff. But that's getting increasingly old-fashioned, and everyone connects on wifi now, which seems to often not be up to the task. We have some huge pipe at the university I work at, and my wired office connection is great, but the wifi? Frequent dropouts and latency spikes, especially if a bunch of people in one room are doing network-intensive stuff. I can believe our IT infrastructure isn't among the best, but it's not the only place I've had that trouble either.

-----


I suspect this is using Mainframe2 which I recall reading about here: http://www.cringely.com/2013/10/15/mainframe2-runs-super-pow...

-----


You are probably spot on, thanks! And nice to see to read/see Mr. Cringely (silicon valley documentary fame).

-----


"If network connectivity is lost, you will need to launch a new session."

As an Australian, I fear this is yet another one of those things we won't be able to use on our spotty 250KB/s connections.

-----


It obviously isn't VNC style. After all, if you lose a connection when using VNC, you don't end up with the files on your local machine to be recovered. They would be still out on the net.

The processing is local, that is for sure.

-----


I don't see pixel-level precision being hard. As for color-correction, I would bet they have settings to allow you to adjust based on your monitor.

-----


Doing color correction based on monitor colors would be impossible: sometimes you adjust, for instance, RGB curves for one specific layer.

It could mean that half of the screen is changing with every step of the slider, and you want exact color, immediately.

-----


I think you're making this overly complicated.

For one, doing client-side color correction (for example with WebGL) is technically possible. It's probably not in this solution, but let's not use words like "impossible" unless they're actually impossible.

But more importantly, I think you're just complaining about latency, but then you're using the phrase "impossible."

EDIT: I thought you were talking about gamma correction and the color space of the display, where the program running on the server wouldn't know about the gamma and color characteristics of the monitor on your current browser. Thus, "monitor settings." But no, you're just worried in general about whole-screen fill rate latency. Again, I think using terms like "impossible" is hyperbole. It will be fine for some users, and not for others.

-----

More

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact

Search: