Open qcha0s opened 6 years ago
I'm seeing the same issue - no synchronization between windows. Trellis works for me, but the mpl demo instructions from README don't work.
I think the Electron log error is benign though, I also get that in Trellis.
I've tried copying some of the settings from Trellis, which does work, but that hasn't helped.
The settings I copied are:
MPL.config.name
process.env.PORT
webPreferences: { experimentalFeatures: true }
to the BrowserWindow
preferencesI also tried changing the package.json
dependencies to match Trellis exactly.
Is there something about the subclass of MPL Store in Trellis, or some other setting, which is needed to get things working as the README describes? Thanks!
Hmm, it has been a while since I've run Trellis. Might be worth trying it out again. I think I'd recommend probably building on hypermerge instead of Trellis. The signalling server requirements for WebRTC (and therefore MPL) turn out to be a real pain for practical day-to-day usage.
I've tried a fresh app using electron-forge and the sample code provided in the readme, and I'm getting the same issue. It's weird that Trellis is working fine but that a simple app is not.
Some things I've tried:
process.env.PORT
to be 4242
like TrellisI'm confused at how Trellis is able to sync/connect but a simpler app cannot.
Oh, it might be the firewall rules for your OS. Are you on Windows? Mac?
I'm on Mac OS. When Trellis runs (I'm in dev mode on both Trellis and my own Electron app) there seems to be two OS dialog boxes that open asking for permission for the app to use the network. On my Electron app I don't get that OS dialog. I've compared code so many times, but still can't seem to find a difference. It makes me think that somehow MPL isn't starting a network connection. Trying to find other things to investigate.
On Fri, May 24, 2019 at 5:54 PM Peter van Hardenberg < notifications@github.com> wrote:
Oh, it might be the firewall rules for your OS. Are you on Windows? Mac?
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/automerge/mpl/issues/13?email_source=notifications&email_token=AAAABAJOAB7IDGXFON3ZDDTPXB54LA5CNFSM4ERT6UU2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWGY24Q#issuecomment-495816050, or mute the thread https://github.com/notifications/unsubscribe-auth/AAAABALUYNUZRC7EPJX27UDPXB54LANCNFSM4ERT6UUQ .
Is there a reason you're building on MPL instead of hypermerge? From my perspective, Hypermerge is quite a bit more mature and robust but maybe it has some properties you don't like? I'm open to doing some investigation/maintenance here if there's a use-case I just haven't grokked.
It was mostly ignorance of hypermerge. I'm building a spike based on a fork of PushPin now which is going well. The only thing that made MPL more appealing was (a) documentation, and (b) slightly less complex dependencies, e.g. I was having trouble with libsodium and another native lib that I can't recall just now.
On Sun, May 26, 2019 at 11:39 AM Peter van Hardenberg < notifications@github.com> wrote:
Is there a reason you're building on MPL instead of hypermerge? From my perspective, Hypermerge is quite a bit more mature and robust but maybe it has some properties you don't like? I'm open to doing some investigation/maintenance here if there's a use-case I just haven't grokked.
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/automerge/mpl/issues/13?email_source=notifications&email_token=AAAABAPIIYLY4VQ56ENGKFDPXLDMJA5CNFSM4ERT6UU2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODWIKFWY#issuecomment-496018139, or mute the thread https://github.com/notifications/unsubscribe-auth/AAAABANDA2CCU7UOVFXYWBTPXLDMJANCNFSM4ERT6UUQ .
[21426:0220/213228.333069:ERROR:CONSOLE(7323)] "Extension server error: Object not found:", source: chrome-devtools://devtools/bundled/inspector.js (7323)