Tom Critchlow wants to run River5 on Glitch

[scripting]

Tom Critchlow wants to run River5 on Glitch. Has anyone had any success with this?

https://twitter.com/tomcritchlow/status/988452117823348736

Dave

[SparkyT]

Yes to some extent :-) - http://newriver5.glitch.me/

Steps:

• Start new glitch instance
• Open the advanced options for the project, grant access to allow import/export
• Import from GitHub (scripting/river5)
• Go to config.json
  • Change config.json so you have extra lines as follows:
    "httpPort": 3000,
    "flDownloadPodcasts": false,
    
    httpPort sets the port that Glitch likes so you don't need to fiddle anymore
    flDownloadPodcasts stops Glitch downloading all of the podcasts in the example feeds which can quickly fill up your Glitch instance
• Wait for it to rebuild
• Click on the "Show - Live" button, this jumps you to an https version of the Glitch instance (e.g. https://newriver5.glitch.me/). Due to there still being some references to the http version of fargo.io in the repo, change the address to be http instead (e.g. http://newriver5.glitch.me/). This should then show you what you want
• You can then add new feeds into the relevant folder & have then show by editing config.json to add a new tab
• Alternatively, just remix newriver5.glitch.me :-)

[tomcritchlow]

@SparkyT amazing! Thank you. I'm gonna check this out. Thanks

[scripting]

Questions for @SparkyT --

0. Thanks for helping us out. That's the first thing I want to say. Now the questions.

1. Someone told me that background tasks don't work on Glitch. Not sure what this means. River5 couldn't work without being able to set up some calls using setInterval.

2. It looks like your server reset in the last hour or so, started reading feeds from scratch? Not sure if you restarted it, or something went wrong. Have you been using this River5 installation or is it something you set up as an experiment?

Dave

[scripting]

Looking at the dashboard I can answer one question --

http://newriver5.glitch.me/dashboard

The server has been up for 2.1 hours. :-)

Dave

[SparkyT]

Isn't it always the way, you have something running nicely, share it and it all breaks :-( ?

I'm not sure how Glitch deals with the attempts to run continuously but it does go to sleep after a while and then completely rebuild/recreate the feeds when I go back to it.

It appears to have filled up the assigned space on the newriver5 instance (maybe due to it being up longer than my tests previously made happen). Getting it back up and working seems to require some wrangling which I will attempt now.

[scripting]

Thanks — let’s figure out what’s needed to keep it running smoothly. Maybe we could have the client call back periodically to keep it running.

[scripting]

I added a one-line script here that loads a quick page on your server every five minutes.

http://newriver5.glitch.me/now

[SparkyT]

Timezones meant I had to leave the conversation but the ride to work this morning triggered a memory or two

• Glitch instances sleep after 2 minutes of inactivity and are destroyed (whatever that means) after 15 according to this post https://support.glitch.com/t/idle-time-before-sleep/376/4
  • I'm not sure what that means given that hitting the instance with an http request seems to keep it up.
• In the past for testing I've user Uptime Robot (https://uptimerobot.com/) to do the pinging for me
• Long term, Glitch probably isn't useful as anything than a testing ground for River5 mainly due to the issue above and the fact each instance has 128MB of storage (node modules don't count against this) so a large number of active feeds or having flDownloadPodcasts set to true will rapidly fill it up and leave you scrabbling around trying to get it back working again.
  • I've added in the feedBase.io top 100 feeds as an example and will now stop playing with this instance so it can be a reference that doesn't keep breaking :-)

[scripting]

Yes, I came to the same conclusion.

http://scripting.com/2018/04/24/124932.html

I like the way you used the feedBase top 100. Very inventive. ;-)

Dave

[cori]

I don't know anything about the configuration available for River5, but if the app's data could be stored in .data instead of in data, glitch maintains the content of that folder when the container is restarted, and it's also not copied on remix (so a remix wouldn't start out with someone else's River5 data folder contents).

[scripting]

Cori, that makes perfect sense, and River5 is totally configurable this way.

Here are the docs for config.json --

https://github.com/scripting/river5/blob/master/docs/CONFIG.md

If @SparkyT is listening --

All you should have to do is change the value of dataFolder in config.json to have the value ".data/".

All the data folders that River5 creates or looks for are in that folder.

Perfect. I like the way the designers at Glitch think.

Dave

[cori]

Great! I thought it might be but didn't see it in newriver5.glitch.com's config.json at a quick glance.

For reference there have been a number of efforts to port River5 to Glitch previously: https://glitch.com/search?q=river5

[scripting]

Cori, that means he's accepting the default. You only have to put things in config.json if you want to override the defult.

Adding this one line to his config.json should fix this problem --

"dataFolder": ".data/",

That should cause it to create all its data in a sub-folder of the Radio5 folder called .data

Dave

[SparkyT]

Thanks all, dataFolder change made and stuff is being written there.

[scripting]

The app seems to be gone now.

I still wonder if what it's doing in the background is kosher.

I would search for setInterval in the Glitch docs and see if it has any problems with running code from those threads, or what happens to them when they shut off an app.

It could be they shut River5 down when it's in the middle of writing a file. That could definitely screw things up.