Really cool so far. However, when I work for some time in the editor, it tends to throw me this error every 5-10 minutes. I then usually have to wait 2 minutes for it to come back.
Hmm… We’re having trouble connecting. Try refreshing?
I don’t remember having this problem with my previous project (at least, not as often). The only big difference is that this one uses websockets (in this case socket.io)
Is this a normal behaviour? Is there something I could do on my side to reduce the frequency of this problem?
Sorry to hear that. Certainly isn’t normal behaviour - I’ve seen it from time to time myself but certainly not for that long or frequently. Does refreshing your browser tab help? We’re working on some major backend changes, which we hope will help with issues such as this, but we’re a few weeks from having them available. In the meantime, if I can get the project name then I can try to reproduce it. Thanks.
Hi there, I have been having this problem quite often on my own project (bloom-thief). Some of the worst occurence involved the page being frozen for a couple seconds and then my work reverting to a previous state.
I am using Chrome on Windows, and thought some of my extensions might be causing problems, namely “Https everywhere” and “Disconnect”. Disabling them helped somewhat, reducing the occurence to every 20mn instead of every 5mn.
Sorry to hear that @rcsl! These issues are usually server-specific, so as a temporary workaround you can try remixing your project and hopefully it’ll be on a different server. Meanwhile, I’ll get the team to kick the servers and our long-planned server improvements are almost at completion, which should provide a more conclusive resolution to such problems. I’ll let you know when we have something to announce on that front.
I experience the same issue, also with a socket.io related project. In it’s current state, the environment is pretty much unusable. It allows me to edit for a couple of minutes, then crashes and comes up after aprox. 3 minutes, repeating the cycle.
There is a rather simple way to solve this for the user without fixing upstream libraries or even the doing very much on the server side.
The GUI and the connection to the server should not live in the same window, because you only really need to reload the connection window to reestablish the connection and the GUI can carry on as normal, without the user having to step away.
Basically, the GUI should load in the top frame, and the page that holds the socket to the server should be in an iframe
inside it. When connection is lost, only the iframe should reload.
Meanwhile, the GUI page should store its state on the client (probably IndexedDB), so that it can be synced with the backend when it gets reconnected, even if the user has closed the tab in the meantime.
sorry to hear that! can you give us the name of the project you’re working on? Lately it has been very rare, and a transient situation… if you get it consistently, we will immediately investigate!
Saw another reset since your message; made the project public. Was still happening every 1-2 minutes; now hasn’t happened in 15 minutes.
Site doesn’t seem to be publishing anymore, so going to reboot.
Ok, the “publishing” problem was just a side effect of the “trouble connecting” issue. The changes I had made (in neutral.js) weren’t saved, so I saw an old version when I went to the web site. When I rebooted, the changes weren’t there; when I re-made them the correct version appeared.
It doesn’t seem to be a problem with my connection, since I was able to load new pages in the browser even while the work wasn’t being saved in the gomix IDE.
I’m perfectly prepared to believe that this happens more often with a poor connection, since it does seem to happen more often in certain locations. That said, the connection isn’t terrible:
Gomix always immediately reconnects within 1 second of showing the “Hmm …” message
I can use other active server-backed web services and games without seeing anything similar
It seems that, when there is a momentary connection glitch, the choice is being made to throw out work since the last autosave. It is technically possible (other sites do it) not to throw out the work, but to simply retry and resume saving once the connection is re-established.
This is unfortunate may I ask you some additional questions?
how often do you see this behavior? We deploy few times a day (0-10, depending on the day…), and during the deploy the editor disconnects briefly. If this is the range of times you see the disconnections, it’s unlikely that you’re experiencing server instability. On the other hand, if it happens more often, we will look into it immediately!
offline editing is on our to-do list. We understand that it is very desiderable, and we are figuring out ways of implementing it in a stable way. However, at the current time the “content reset” should be very very small, in the order of seconds of work. What’s your experience? How much time of work you see reset when a disconnection happen?
I see the issue every few minutes. I generally lose between 5s - 1min of code, plus it resets to the top of the file. I have seen a few cases of larger losses.
Last time I was running ping gomix.com (pinging every second) in a different window while I was working. I saw the issue as usual, and my ping times ranged from 63.8 to 93.8 ms; there were no excessive ping times.
may I ask you some additional questions?