VNC Client

Don’t know about RISC OS, but I use RealVNC’s CNC Viewer with my PC.

https://play.google.com/store/apps/details?id=com.realvnc.viewer.android

I also use RealVNC on my Android tablet, and it connects to RISC OS happily, at least on my LAN. Even caters for 3-button mice!

TightVNC working well her on Windows10.

V0.24 is stable on my ARMX6, no crashes in two days – thanks James.

Windows 10 using TightVNC server is excellent. Being able to view the tiny Dell laptop screen on my large 4k RISC OS screen is a big improvement.

Not everything in the garden is rosy however.

trying to control a Mac Mini (M1) using its built in VNC server is painful. This is not the fault of Avalanche, but to do with poor quality code and bad design of Big Sur Mac OS.

One problem which might be Avalanche related is that there is no visible cursor when I first logon. The Mac requires its own logon before it allows access through VNC. At this stage there is no cursor visible and I have to guess where the cursor might be in order to choose a user profile.
After logon there is still no cursor until I can start an application – say Safari. At this point a pointer appears.
My Avalanche settings are the same as for Win10 which has no such problem.

I can release a new version which allows the session title to be specified.

Yes, please.

How did I miss this earlier? Sorry.

Sadly here on my ARMX6 the new version of Avalanche v0.24 seems to work but keeps disappearing while in use on viewing my PC (Win10) both the window and and icon disappear seemingly at random. No error messages that I can see.

Chris – what VNC server are you using on the Windows machine?

and what level of RISC OS – I am using 5.29 on ARMX6.

On the desktop PC (Win10 Pro) I am using UltraVNC and on the RISCOS 5.29 on ARMX6. never had any crashes when I used v0.22 of Avalanche.

I do have TightVNC on my Win10 Pro Laptop, will see how that goes when I get a minute.

Chris – it is worth a try to eliminate a variable. My ARMX6 has been running Avalanche to Windows10 continuously for 7 hours with no problems so far.

Thanks to all of you for the feedback on Avalanche. I’d like to get a generally stable release before looking at any additions.

It it compiled for general compatibility, so no neon, vfp, etc. There isn’t much fp in there in anycase.

Often I can’t reproduce these issues locally. I’m currently running it on a rpi.

What I found useful locally for locating issues where Avalanche crashes or just exits were the debugger diagnostic dumps, or ZeroPain logs. I had a crude diagnostic module way back which sat on the error vector and attempted to get a backtrack, but the debugger log is far better.

Avalanche does log to SysLog, that can help sometimes.

If Avalanche can’t connect or the connection drops then you should get a message popup to say why. If not then I’ll probably have to create a binary with a load of targeted logging enabled. I could improve this by at least recording roughly what was going on when the disconnect happened.

I’d like to get a generally stable release before looking at any additions.

When you get round to adding session name labelling in the client, might I suggest that you use the network name lookup facilities: DNS and, as a fallback, Hosts.

I have some errors when trying to copy something inside my spreadsheet.

After a few big copy operations on the client side, I have “Message protocol failed” for each copy operation (no error on paste), until I shut down Avalanche. I’ll need to make some more tests with the new version.

When you get round to adding session name labelling in the client,

I still think this is a problem for the server, not the client.

Jeffrey has now offered to add the session name to vnc-server.

I still think this is a problem for the server, not the client.

Well, if the Client displays the name of the device that the VNC server is running on which it derives from a DNS lookup (PTR) you get the name (or IP in the absence of a DNS record or hosts entry¹) of the remote device (server) displayed in the client.

¹ Some people round here don’t seem to like DNS and use hosts files instead.

After a few big copy operations on the client side, I have “Message protocol failed” for each copy operation (no error on paste), until I shut down Avalanche. I’ll need to make some more tests with the new version.

Interesting, a couple of questions:

• How long is the text getting copied?
• What is showing the “Message protocol failed” error?

Looking at this, with the currently implementation Avalanche won’t currently be able to send a selection to the server if it won’t fit in the VNC output buffer (32KB). In that case you should see a message similar to “Unable to send selection to server, too long (… byte(s)).” briefly in the status bar at the bottom of the window. Currently Avalanche won’t attempt to stream the selected text out or hold it until there is space in the buffer so there could be some improvement made there if pasting longer blocks of text is needed.

However… I suspect it probably isn’t getting that far. Tried dragging in just under 32KB of *h.. and that didn’t work as expected, so that may be related.

How long is the text getting copied?

No way to know. Sometimes, when you copy data from complex programs as spreadsheets it copies value, and sometimes value and a lot of formatting information (this is very true for Fireworkz). Anyway, there is no need of a a bunch of data.

What is showing the “Message protocol failed” error?

There is no name in the window. But as soon as I quit Avalanche I don’t have this message any more. So I suspect the message is coming from Avalanche. Of course, when I make a copy and have this message, I can paste the data in the client, but not on the remote system.

when you copy data from complex programs as spreadsheets it copies value, and sometimes value and a lot of formatting information

Ugh. Spreadsheets. The other day I created one in Google Sheets for trying to record Tassimo barcodes and what happens (to hopefully crack the bit assignments). Somehow, towards the end, the thing became “View only” and all attempts at editing were blocked. Neither the app nor the terminally slow web version offered any clues as to how to undo that. In the end, I copied the entire thing and deleted the original (ooh, the Git rescue solution!) and was able to carry on editing. But my hat off to Google for making it easy for a spreadsheet owner to somehow lock themselves out of editing their own spreadsheet, and damn near impossible to undo it.
Of course, if Google had any real documentation that was neither forums or help topics aimed at people who don’t know what a spreadsheet is… the only thing I found out was “this lockout thing happens” followed by a half dozen solutions, none of which worked. Par for the course, that.

So, yeah, spreadsheets. Grrr.

So, yeah, spreadsheets. Grrr.

Fireworkz under RISC OS works fine for me! Perhaps you made a bad choice!

Perhaps you made a bad choice!

Let me know when I can edit a Fireworkz sheet on whatever device I happen to have handy, and then embed the thing into my blog with it being able to automatically update to reflect changes (so no static HTML).

ik.me accounts come with a small kDrive space. And kDrive can edit Office files (Thanks to the OnlyOffice backend).

Fixed a bug which could have caused disconnects or potentially crashes if a lot of data destined for the server backs up. This shouldn’t happen often, but large clipboards could be a trigger. This may well include the spreadsheet problem from what you’ve said above. I’ve put up a new release: https://github.com/effarig/ro_avalanche/releases/download/v0.25/avalanche.zip

Thanks James. I’ll test it this week-end.

I still have the “Message protocol failed”, sometimes, when copying things in Schema2 (local side). The error disappears as soon as Avalanche is closed. Else, it’s really stable.

David, thanks for testing that again.

I still have the “Message protocol failed”, sometimes, when copying things in Schema2 (local side).

Installed Schema2 (1.24 (28-Feb-2021)) and did a lot of CTRL-C, with individual cells and blocks of cells and haven’t managed reproduce this yet. I there any particular operation which is particularly good at inducing the problem.

I suspect that error is reported by Schema as Avalanche doesn’t have that message. When Avalanche is sharing the clipboard, then when any other application claims the clipboard by doing CTRL-C say, Avalanche immediately requests the contents. So what I suspect is happening is that there is some disagreement between Schema2 and Avalanche over the wimp clipboard or data transfer protocol.

I make a lot of local and remote copies, and after some minutes, the problem appears.
Perhaps it’s linked to the fact that I have also some RDPClient session running.