SSH connections fail
Default is now to not use SFTP, closing.
Locks up after failing to SSH-authentication with SYSPASS
Should be fixed in git now.
Hrm, I can't seem to repro here... was this with the nightly or with an MSVC build? If an MSVC build, I wonder if cryptlib doesn't have all the patches... I do note that the .lib hasn't been updated, but from what little I know about link libraries, that shouldn't make a difference.
Hrm, I can't seem to repro here... was this with the nightly or with an MSVC build? If an MSVC build, I wonder if cryptlib doesn't have all the patches... I do not that the .lib hasn't been updated, but from what little I know about link libraries, that shouldn't make a difference.
Mouse selection doesn't work right in full screen mode, Windows
Should be fixed in git now.
To further expand the matrix, External scaling vs. internal (ie: blocky or pointy) also has different behaviour.
Mouse selection doesn't work right in full screen mode, Windows
Note that the second video is in SDL mode which has a different bug with gutters.
Ok, so it looks like you're always trying to start the selection in the "gutter" to the left of the text when in full-screen... are you able to select by starting with the mouse on top of the first character you want to select (as you do in windowed mode)?
The video does not have the mouse pointer in it, so it's hard to tell exactly what's happening.
Blackbars on sides after transitioning from X11 Fullscreen to X11
Alt-B (View Scrollback) from main menu frequently crashes in Windows (GDI) mode
Windows build doesn't run via explorer
Presumed fixed for over a week, closing as fixed.
Rats. Once I get a mac back up and running, I'll rummage around in the SDL APIs some more.
I've update the SDL library version and bumped the required macOS version to 10.7 for the nightly build now. This may do the trick.
Yeah, I'm trying to work out what the logistics of something like that would be... it has to detect "the right kind" of early disconnect since many BBSs will disconnect right at startup due to all nodes in use, zone mail hour, door maintenance, etc, so a message like that needs to not be wrong. The other thing I'm considering is a different connection type... SSH and SSH with SFTP... which is something I've done for other stuff in the past.
What a wild ride this one has been... So there was a few issues that were causing SSH hangs with your BBS and I've fixed them all (I think)! The reason the SFTP is not working is that it appears Mystic will accept any extra SSH channel request, but as soon as any data arrives on another channel, it will close the whole connection. Ideally, Mystic would either refuse the channel request, or it would close the channel as soon as it's opened... Synchronet has historically done the second (because Cryptlib,...
SSH connections fail
So, with SFTP disabled, I still get a hang, it just happens later... digging in, it appears that my side is receiving an invalid SSH packet (the size field is larger than the max allowed). Usually when I see that, it means something else is writing to the socket, but I can't pin down anything that is clearly suspicious yet.
Aha, when I connect to the right port and not port 22, I can reproduce an error (it's a bit different, I do see the BIOS screen rather than a blue screen), but I can see it's from the SFTP channel. I'll update the ticket when I have this resolved.
I would prefer to actually fix whatever the issue is if possible... the option is much less usefull if it's not on by default. I'm trying to connect to The Quantum Wormhole to test, but I'm having a bunch of other issues instead of the one being described (I can't connect with OpenSSH either)... can you give me the address of a BBS this is showing the problem so I can reproduce it?
Thanks for opening a new issue! Part of the problem with continuing old issues is that an email is sent to everyone who has posted in the past. So first off, the warnings are not a problem... despite the code explicitly ignoring the values (with the (void) cast), the compiler is still warning because the author of the function believes that you should always care about the return value... in these three cases, that is false. Next up, regarding libncursesw.so.6, this is also what I'm using on my systems....
SSH connections fail
The bug that was reported (SSH connections freeze after entering the system password on windows in GDI mode) has been fixed and no longer occurs for the reporter. You're talking about what is clearly a different issue (SSH connections from Linux in ??? mode hang at a blue screen when connecting to ???). The issue you're seeing does not occur on the Linux, FreeBSD, or Windows systems I test with when connecting with my BBS or Vertrauen (both of which offer SSH logins), and has apparently not been...
It turns out, the Windows system I was testing on had enabled the Fault Tolerant Heap and hidden this issue from me. Should be fixed now.
It turns out, the Windows system I was testing on had enabled the [Fault Tolerant Head|https://learn.microsoft.com/en-us/windows/win32/win7appqual/fault-tolerant-heap] and hidden this issue from me. Should be fixed now.
ALT-U no longer working
Blackbars on sides after transitioning from X11 Fullscreen to X11
Ok, now this should be fixed... if it isn't, there's a reasonable change I'll wont-fix this one.
Please note, this bug has been closed, so is no longer being tracked for the upcoming release. If this is still an issue, please open a new bug with as much information as possible to reproduce the problem. As for the warnings, they are harmless.
Here's what's happening... 1) When exiting in fullscreen mode, the scaling is stored in syncterm.ini. 2) When SyncTERM is restarted in window mode, the scaling from syncterm.ini is used. 3) The WM in ChromeOS does not support the _NET_WM_DESKTOP atom, so SyncTERM only knows the size of the screen, not the size of the usable desktop area... as a result, SyncTERM creates a window of the same size as the screen. 4) The WM resizes (and usually moves) the window to avoid overlapping the panel. 5) SyncTERM...
Blackbars on sides after transitioning from X11 Fullscreen to X11
What BBS are you trying to connect to? May also be useful to get the git revision you're building from.
Blackbars on sides after transitioning from X11 Fullscreen to X11
Still unable to reproduce this on any of my X11 servers (FreeBSD, ChromeOS, Ubuntu 20, and MobaXTerm under Windows).
So my guess on this is that SyncTERM isn't managing to get the actual real resolution from macOS. macOS very pervasively works to make it hard to actually do things with real pixels, which is what SyncTERM is trying to do. Updates to libsdl keep getting better at this, but Apple also works to make it harder at the same time.
Option to disable mouse support, per BBS entry
Added ALT-O to toggle remote mouse support.
If the entries do not have a specific path configured, you can change the default via the "Default Connection Settings" menu.
Blackbars on sides after transitioning from X11 Fullscreen to X11
Still unable to reproduce this on any of my X11 servers (FreeBSD, ChromeOS, Ubuntu 20, and MobaXTerm under Windows).
Alt-B (View Scrollback) from main menu frequently crashes in Windows (GDI) mode
Wow, not at all what I understood from the description... video was very helpful. I can't reproduce this with the just-rebuilt nightly now (but have never reproduced it), so it may have been fixed with the rect validation change.
Unable to reproduce with the just-built nightly. May have been fixed with the rect validation fix, but I've never reproduced it on this Windows install. Seemed to fail about one in six times with the VM (which I'll test later tonight).
SSH connections fail
ALT-U no longer working
This appears to be working in the currently nightly.
ALT-F4 doesn't work in GDI mode
Fixed with ce7c2cca9
SSH connections fail
Unable to reproduce with the current nightly and cvs.synchro.net.
Windows build doesn't run via explorer
I'm not able to reproduce this with Windows 11. While I do get a warning that it's from an unknown publisher, if I select run anyway, the icon in the toolbar and in Explorer is correct, and it launches and pins to the toolbar correctly. No ghost processes appear in Task Manager (plist doesn't appear to be part of Windows).
The issue here is that comio doesn't presently have a way to query what the serial ports support, and the APIs to use these higher speeds tend to be poorly defined. At present I'm fairly limited in what I can test, so guessing APIs and hoping it works is where I'm mostly at right now.
This is expected behaviour. While curses and win32cio ciolib modes allow "batching" updates so the cursor doesn't end a draw in the status bar, graphics modes use a timer to snapshot the screen state and there's no way to ensure the cursor isn't drawn in any specific location during a cputs() or cprintf() as is used in the status bar. SyncTERM could use puttext() to avoid that, and ciolib could get an update transaction thing, but these would be new features.
Ticket moved from /p/syncterm/tickets/130/ Can't be converted: _output_type: Win32-GDI _found_in_version: 1.2b
Possible fix committed as 736d4169
Another project that has struggled with this: https://www.lexaloffle.com/bbs/?tid=29705
ALT-U no longer working
ALT-F4 doesn't work in GDI mode
Alt-B (View Scrollback) from main menu frequently crashes in Windows (GDI) mode
Windows build doesn't run via explorer
SSH connections fail
Cursor visible in status bar while updating/redrawing
syncterm 1.1 no longer compiles
Closing this as won't-fix. This compile fails because of a change made to the C library after 1.1 was released. The referenced patch will do the trick.
Ticket moved from /p/syncterm/tickets/120/ Can't be converted: _milestone: 1.2
Ticket moved from /p/syncterm/tickets/131/ Can't be converted: _output_type: SDL _found_in_version: 1.2b
Ticket moved from /p/syncterm/tickets/132/ Can't be converted: _milestone: 2.0
Build with cmake failing
Build instructions for SyncTERM are in src/syncterm/COMPILING.
Ticket moved from /p/syncterm/tickets/127/ Can't be converted: _output_type: SDL _found_in_version: 1.0
I was replying to @rswindell regarding the scrollback buffer, not the highlight bar in menus.
ALT-F4 doesn't work in GDI mode
I really don't understand what you're saying... in everything on every OS I've ever seen, if you push the scroll wheel away from you one click (ie: scroll up), a line of content in the scrolling region moves to a lower position in the region. Scrolling up moves the view of that data closer to the beginning. This is what SyncTERM does in the scrollback.
I really don't understand what you're saying... in everything on every OS I've ever seen, if you push the scroll wheel away from you one click (ie: scroll up), a line of content in the scrolling region moves to a lower position in the region. Scrolling up moves the view of that data closer to the beginning.
Cannot type character with hex code 0xe0 ('р' in CP866 or 'Ю' in KOI8-U)
Cannot type character with hex code 0xe0 ('р' in CP866 or 'Ю' in KOI8-U)
This should now be fixed in the current git build. I've updated the nightly build on the website as well. Please let me know if this works for you.
Download request for 1.2a for Linux
Closed at reporters request.
The "current" connection is a copy of the one in the dialing directory, and isn't updated when the dialing directory is updated. While a small number of settings could be applied to the current session (ie: passwords), most of them (connection type, etc.) can't be, so that functionality was never added.
System password not sent with ALT-L if added after connection
Ticket moved from /p/syncterm/tickets/122/ Can't be converted: _milestone: 1.2
Generally, that's asked for ssh connections. SSH requires a UserID and Password to make a connection. For Synchronet BBSs, you can enter the user ID you would like to have (password won't matter) to create a new account on the BBS. Other BBS packages may require you to log on with a specific UserID to create a new account, and still other BBS packages require everyone to log in as the same user (often "bbs") then do a separate login.
So, the issue here is that SyncTERM uses the ciolib library, which is an implementation of the old Borland conio for DOS. The getch() function returns a scan code, not a character, and for scan codes, 0xe0 indicates it's an extended two-byte scan code. It looks like I'll need to "double" 0xe0 entries in every input style, then add code in everything that calls getch() to handle 0xe0e0 correctly and fixup the unicode translation tables. I'll try to get this hacked up this weekend, but I'll need to...
Try in SDL mode. If it works in SDL mode, but not in GDI mode, please file a bug. This was fixed before the GDI output was implemented, so it's possible that it never worked for GDI.
GDI mode is the default for Windows, so it should work. You can change the mode from the config or the command-line, but I'd rather fix the bug. :D I expect it should be an easy fix now that I'll look in the right place.
Use alternate dialing directory
syncterm -b<altfilename.lst>
So, looking at the ini file, it's not a SyncTERM setting. One thing I don't see though is a mode selection, are you sure it's SDL mode? It should say in the top-left corner of the main window. With the mention of APPDATA, it sounds like it's on Windows, and the default for Windows has changed to GDI mode.
Another option is to run the command: konsoleprofile "colors=linux-colors" in Konsole before launching SyncTERM.
Arrow keys printing characters once in BBS
So, my install didn't find libcurses-dev, so I used libncurses-dev. You shouldn't need to set SRC_ROOT. It's a good idea to install pkgconfig as well (though it's not needed) So, I've fixed it, but Konsole has screwed up the standard colours. For SyncTERM to look correct in Konsole, you need to create a new profile, and in "Appearance", set the Color scheme to "Linux Colors" then switch to that profile.
So, no issues in Konsole here (except the colours look terrible). I'm going to a wedding this weekend, but when I get back, I'll try installing Kubuntu in a VM and see if I can repro.
And it shows "Curses" in the top-left corner when at the main screen? I'm truly mystified. I'll try it in Konsole here (on FreeBSD) and see if I can reproduce it. I'm not aware of any terminal that would send those characters for arrow keys, and I can't think of any way ncurses would map them like that. I was really expecting an ESC sequence with a weird terminator, not a bar character (though you've possibly filtered those out) with the grep. You can attach files here.