Project broke glitch and all my browsers


#1

Hey guys,

i did “a stupid” thing which “broke” glitch in my case! :wink:

i implemented the lib twit for twitter streams and it works as expected but as just a few tweets came on my hashtag i decided to use “trump” and now i can not access my app anymore because it has to much debugging output to handle.
all of my browser collapse if i open the project, and i can not change just this one line so that it will work again.
i can not even delete the project because i can just do this inside the edit mode which again … let all my browsers collapse.

the projects name is “twitter-carousel”.

i would prefer to access the code to change just this one line somehow, but if you guys have to delete it you can even do this.

sorry fpr the inconvenience!

regards, Michael


#2

Might be the planned downtime. Come back in a few minutes.


#3

@drew thank you for your Feedback but it is not.

i console.log(tweet) every tweet-object which comes out of the twitter stream on “trump” and it busts every browser. it is somekind of “endlessloop” where i can not change this line because the editor breaks faster than i can change this line! :joy:

i tried a lot of tricks, including very fast clicking and scrolling on loading but no chance!

regards, michael


#4

Hey @woo.drama,

would reverting the code to a previous time fix this issue?


#5

Hello @etamponi,

sure it was the last thing i could do after i was “locked out” by the browser performance.
how to?
i just need to change one line in the server.js and replace the trump string of get rid of the console.log!

anyways … even if the “log”-interface is closed it somehow shows those serverside console.logs somewhere in the interface hidden i guess. is that smart? just a question i ask myself caused by this problem i have. :wink:


#6

I did it for you :slight_smile: You should be able to edit it now


#7

Hi @woo.drama, not that my reply here is anything to do with your app or Glitch, but interestingly I did a similar thing with a Go program earlier in the year. I ‘tracked’ (Twitter parlance) ‘Trump’ and was processing an average of a little over 100,000 tweets per hour.

The most I ever saw was 172,796 in one hour period. Twitter also told me that it didn’t send me 1,289,084,354 tweets either - yep, that’s over 1 billion tweets with ‘trump’ in it, in one hour. No wonder though that the logging couldn’t keep up. :slight_smile:

(For those interested, this was during the time when Comey was being interviewed after having been fired, hence the elevated number of tweets.)

Perhaps I should retry this experiment on Glitch and see what figures I can get out of it. It’d certainly be interesting.

Hourly Stats

23 Aug 2017, 03:00 UTC

We saw 172796 tweets during this period:

  • 103860 tweets had no URL
  • 20358 had an empty URL
  • 0 had invalid URLs
  • 396 were domain only links
  • 40781 are from domains we ignore
  • 7401 were counted towards the total
  • 2120 were unique URLs

Twitter also told us it didn’t send us 1289084354 tweets at all.


#8

@etamponi thank you a lot! Hopefully i will not trap in something similar so far. But it can happen fast, when you console.log a big Object times … a lot as the trump tweets! :wink:

@chilts i just gave out console.log(tweet) of the raw full tweet object which seems with the frequency the word “trump” is postet on twitter to much for a browser to handle in HTML. Means on glitch.com where socket.io and DOM do have to show a lot of output in my case!


#9

Sounds like we need a way to launch a project into “safe mode” where everything is turned off and suppressed.


#10

alternatively, we shouldn’t even attempt to print logs when there’s more than n amount of them at a time


#11

I think @triptych’s idea is a good one.

@pketh Blindly saying someone shouldn’t console.log() without knowing the repercussions is a little short-sighted (imho). We (a general we, as developers, and I’m including myself) will use console.log() as a tool, and perhaps sometimes we don’t know how many times it’ll be called, which is information we use to debug our apps. If that breaks the platform, then there must be something we should be able to do to get back to a working version.

Of course, there are many ways to do this. Rollback to a certain time, rollback an hour, rollback 5 versions, boot into “safe mode”, stop the app and allow debugging, put the app into maintenance mode. It’ll be up to the Glitch platform team to figure out the best way but hopefully there are some suggestions here, both @triptych’s and mine above, both good and bad, both do-able or impossible. It’ll be interesting to see if you think any of these are worth implementing.


#12

hey @chilts,
my intent with ‘we’ ^, was to mean that the glitch editor client (i work on glitch) shouldn’t attempt to print potentially millions of log messages to the screen. My point is that if the editor didn’t attempt to parse/output every message back, it wouldn’t lock up in the first place.

Apologies for the vague word choice implying that this was something that you should have to consider.


#13

Hi @pketh. Yes, I knew you worked on Glitch, so am glad you’ve cleared up what you meant. Apologies for me jumping in with assumptions as to how I read your response.