New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Neighbourhood view not updated on shared activity stop #840
Comments
|
Was this against a released version of Sugar, top-of-trunk, or a branch? |
|
Yes. |
|
All of the above? |
|
Sure, that's another way of looking at it. At least released versions 0.112, 0.113, 0.114, master, and python3 branches are affected, as are the branches in development for TelepathyGLib porting. |
|
My question might be off-topic. |
|
variant_level is an argument for dbus.Byte, so it is logical by inheritance that it is available for use in a dbus.ByteArray. dbus-python source package, file dbus_bindings/bytes.c is also useful. |
|
Thanks for explaining. :-). |
|
Summary: Debug a Telepathy crash Sugar uses the Telepathy API for collaboration. Every time a collaboration stops, one of the Telepathy programs crashes with a core dump. It has been happening for a few years. See if you can use gdb and the Telepathy source code to find out what the problem is! A task for experts. You may have to reach out to the Telepathy project as well. |
|
@quozl I'm able to reproduce the error in the instance two example differently. Reproduced by: Closing any activity. Error: Tested on: Sugar 117-3, Debian 10.4 Buster |
|
Thanks, but it makes no difference. Telepathy Salut is still crashing, and it should not. It is the Telepathy Salut process on instance two that crashes, hence the set_active message. It is, however, useful to know the same problem persists in 0.117 Debian package 3. Perhaps the same version of Telepathy is used. |
|
I am not sure; I have logs related to |
|
No, it doesn't help, it confuses. "does not support OLPC buddy properties" is a different problem; compare your version of Telepathy Salut, we're using 0.8.1 with Ubuntu 20.04. "NoReply" is the same underlying problem. |
|
for future reference, since I just ran across this issue @srevinsaju ran across back in June, inside the SoaS Fedora nightly image: In the latest Fedora Rawhide (will soon become 34) SoaS images, telepathy-salut is version 0.8.1, and python3-gwebsockets is version 0.7, and metacity is version 3.38.0 |
|
For the completion of my answer that time, I was using Tested again, have the same error. |
|
Indeed. It needs diagnosis by debugging Telepathy, or reviewing Telepathy commits. My guess is the support for Sugar may have been removed from Telepathy because it was not maintained. |
|
This still happens under Sugar 0.118 in the Fedora Rawhide nightly composes. Upon examining the Telepathy source repo for Sugar support, it does does not appear to have been removed from current releases of Telepathy, only telepathy-next. Fedora Rawhide/34 includes telepathy-gabble 0.18.4, telepathy-salut 0.8.1, telepathy-glib 0.24.1, and python3-gwebsockets 0.7 |
|
Thanks. That eliminates my guess. It still needs diagnosis by debugging Telepathy, by testing an older version of Telepathy, or by reviewing Telepathy commits between the last working version and now. To be sure you have the same problem as described by this issue, the Telepathy Salut process id will change between when the shared activity is running, and when it is stopped. If process crash dumps are configured, then a core file will be made. |
|
This also happens when a shared activity has NOT been stopped, so the bug description should be updated to reflect that, since the scope is significantly larger. Here's the log output I get from Sugar 0.118 (SoaS Fedora 34 prerelease): |
|
@aperezbios when a shared activity hasn't been stopped on one of the shared instances? |
|
@chimosky when simply looking at the neighborhood view (with jabber.sugarlabs.org configured as the collaboration server) |
|
Bizarrely, the simple act of setting G_MESSAGES_DEBUG="all" (and rebooting, or killing jarabe.py to induce sugar desktop re-start) in /etc/environment seems to "fix" the problem, although I've no idea why, as it makes no sense. |
The first part of the error could be when no interface was returned which could happen when a connection hasn't been made but that'll throw a I've seen the last line of the error and I can't remember the cause of that and I think there used to be an open issue for it.
Are you saying when you do this the jabber server connection gets created successfully and those errors aren't shown in the logs anymore? |
|
Oh, okay. No, I'm not using jabber.sugarlabs.org at all, I'm concentrating only on link-local peer to peer, aka Telepathy Salut. If you're using jabber.sugarlabs.org, then you're using Telepathy Gabble. Two very different underlying software components. I'm interested to know what causes Salut to unexpectedly terminate, but I've not even got as far as analysing the core dump. |
|
@quozl should we separate these issues, one for telepahthy-gabble, and one for telepathy-salut? We need to figure out why setting G_MESSAGES_DEBUG="all" in /etc/environment (and then restarting jarabe), results in Sugar (using) telepathy-gabble seeing neighbors/peers. |
|
Thanks for asking. I don't mind either way. It's not like were not solving the problem because the issue is too hard to read. Speculating; we're not solving it because (a) it is a Telepathy problem rather than a Sugar problem (most likely), (b) Telepathy is written in C and we have attracted primarily Python developers, and (c) those of us with C skills haven't had the time to investigate. If you do split the issues, create new ones that condense and summarise the observations in each scenario, with a link back to this issue, then we can close this issue. |
When an invited user stops a shared activity, the neighbourhood view may be incorrectly updated.
Reproducer;
Observed behaviour;
Expected behaviour;
Instance one shell.log;
Instance two shell.log;
org.freedesktop.DBus.Error.NoReply is caused by premature termination of the telepathy-salut process before it replies to a message.
On instance two at the time of the event, the process ID of the telepathy-salut process changes, and on Ubuntu 19.04 and an apport crash file is added to
/var/crashon behalf of the process.The problem is always reproducible and repeatable.
Theory; a message sent to Telepathy Salut causes it to crash. It should not crash. This is a bug in Telepathy Salut, possibly combined with a bug in Sugar usage of the API.
The text was updated successfully, but these errors were encountered: