I'm trying to make sense of how one might use vnc via a web browser; evidently it is meant to be possible with packages like 'noVNC'.
The official doc is not exactly fabulous so it's taking me more effort than it really should to see what it meant to happen, but I dare say that will eventually work out. Pointers to a good explanation would be appreciated.
The more directly squeak relevant question is whether anyone has made our RFB server work to display within a web browser. It seems like it ought to be doable, it's just sending bits down sockets after all. The real prize would be finding a way to use the same port that a Seaside connection is using as opposed to having to find a 59XX port that is available, configure port forwarding, and all the other faffing about. You have a connection! Use it you stupid computer! Even doing things the clumsy manual way proved how incredibly useful it could be to get RFB access to a headless server the other side of the continent. Imagine having a convenient way to get that within Seaside. Maybe even imagine making a system that can VNC a single Morph, allowing say, a HierarchyBrowser to be share for a quick debug, or a BlobMorph.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim When flying inverted, remember that down is up and up is expensive
Has anyone else noticed some strange performance issues when using RFB squeak-to-squeak?
If I start a server on one squeak and open a viewer on another it initially opens a very partial view of the server display. Mouse & key presses transmit immediately and the server responds directly BUT nothing gets reflected in the viewer. At least, not for some considerable time, as in "many seconds" and sometimes a minute or more. After that it (almost always) updates the entire view in the viewer and responsiveness is very good - as in you can rapidly drag-select text and it keeps up pretty well. I've currently got a Pi to Pi connection that has been open for 15-16 hours and is quite happy and responsive.
A kind-of counter example though was opening two squeaks on one machine and using 'localhost' as the target machine for the viewer. That test gave immediate full responsiveness. For a while I was wondering if the problem was the bonjour name service, but even using direct IP numbers didn't change anything. I also have recent experience using RFB to debug a headless Seaside system from 3000 miles away with really good responsiveness, though occasional long stutters.
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Objects are closer than they appear.
On Wed, Jun 28, 2023 at 11:35:36AM -0700, Tim Rowledge wrote:
Has anyone else noticed some strange performance issues when using RFB squeak-to-squeak?
If I start a server on one squeak and open a viewer on another it initially opens a very partial view of the server display. Mouse & key presses transmit immediately and the server responds directly BUT nothing gets reflected in the viewer. At least, not for some considerable time, as in "many seconds" and sometimes a minute or more. After that it (almost always) updates the entire view in the viewer and responsiveness is very good - as in you can rapidly drag-select text and it keeps up pretty well. I've currently got a Pi to Pi connection that has been open for 15-16 hours and is quite happy and responsive.
A kind-of counter example though was opening two squeaks on one machine and using 'localhost' as the target machine for the viewer. That test gave immediate full responsiveness. For a while I was wondering if the problem was the bonjour name service, but even using direct IP numbers didn't change anything. I also have recent experience using RFB to debug a headless Seaside system from 3000 miles away with really good responsiveness, though occasional long stutters.
tim
I'm afraid I do not have any experience with RFB squeak-to-squeak, although I can confirm your observation that using it to control a remote headless Squeak server image is really great. It's amazing that it works at all, and even more amazing that it works so well.
Hopefully someone else can speak up with experience using a viewer in Squeak.
Dave
An update -
On 2023-06-27, at 2:58 PM, Tim Rowledge tim@rowledge.org wrote:
I'm trying to make sense of how one might use vnc via a web browser; evidently it is meant to be possible with packages like 'noVNC'.
The official doc is not exactly fabulous so it's taking me more effort than it really should to see what it meant to happen, but I dare say that will eventually work out. Pointers to a good explanation would be appreciated.
Trying to install on my xubuntu-box using the "Quick Start" suggestion of 'snap' didn't result in anything that made any sense to me. Possibly someone familiar with whatever 'snap' does would be happier. So I downloaded the github zip file and did things a little more manually with
cd to the utils folder & `./novnc_proxy` cd to the create websockify folder and `sudo python3 setup.py install` (I tried without sudo and it whined pathetically at me) cd back to utils and `./novnc_proxy --vnc localhost:5902` to connect to the port my Squeak RFB is serving
That provided http://gravious:6080/vnc.html?host=gravious&port=6080 ... which tried to connect but didn't ever complete. When I cancelled it the RFB log output in the Transcript showed the socket being closed, so evidently things got at least partway connected. I've had simialr results when trying to use 'normal' VNC clients to connect to RFB as well and I think I recall the error being something about mismatched protocols?
Connecting instead to port 5901, served by not-Squeak, worked, so it seems to have installed ok eventually. Baby steps...
tim -- tim Rowledge; tim@rowledge.org; http://www.rowledge.org/tim Strange OpCodes: FCE: Fill Core with Epoxy
squeak-dev@lists.squeakfoundation.org