[twitter-dev] Re: alert() in anywhere.js
We just rolled out @anywhere yesterday and some of our users are experiencing similar issues. http://twitpic.com/1p00d6 Steve On May 14, 6:57 pm, Larry la...@topsy.com wrote: I just came across a coworker's browser that triggered analert() call fromanywhere.js. While okay for development, the use ofalert() is not friendly for production websites. Could these be converted console.log() or some other benign mechanism? Grepping throughanywhere.js I found two instances ofalert(): alert(To set up @anywhere, please provide a client ID); alert(No version matching +Z); Cheers Larry
Re: [twitter-dev] Re: alert() in anywhere.js
On Wed, May 19, 2010 at 11:27 AM, Steve C st...@twitpic.com wrote: We just rolled out @anywhere yesterday and some of our users are experiencing similar issues. http://twitpic.com/1p00d6 We rolled out a fix at the weekend that we fixed all the browsers that we test under but there are obviously still some browsers getting the issue. I think we'll use console.info to display these message instead of an alert. We wanted to let developers know that they needed a clientID in the most noticable way but to avoid unintended annoyance of users we'll move to console.log. Thanks, -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong +1 415 425 5631
[twitter-dev] Re: alert() in anywhere.js
What is your ETA on rolling out the change? We are deciding on whether we should disable @anywhere until alert() is removed. On May 19, 2:41 pm, Dan Webb d...@twitter.com wrote: On Wed, May 19, 2010 at 11:27 AM, Steve C st...@twitpic.com wrote: We just rolled out @anywhere yesterday and some of our users are experiencing similar issues. http://twitpic.com/1p00d6 We rolled out a fix at the weekend that we fixed all the browsers that we test under but there are obviously still some browsers getting the issue. I think we'll use console.info to display these message instead of an alert. We wanted to let developers know that they needed a clientID in the most noticable way but to avoid unintended annoyance of users we'll move to console.log. Thanks, -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong +1 415 425 5631
Re: [twitter-dev] Re: alert() in anywhere.js
On Wed, May 19, 2010 at 1:41 PM, Dan Webb d...@twitter.com wrote: On Wed, May 19, 2010 at 11:27 AM, Steve C st...@twitpic.com wrote: We just rolled out @anywhere yesterday and some of our users are experiencing similar issues. http://twitpic.com/1p00d6 We rolled out a fix at the weekend that we fixed all the browsers that we test under but there are obviously still some browsers getting the issue. I think we'll use console.info to display these message instead of an alert. We wanted to let developers know that they needed a clientID in the most noticable way but to avoid unintended annoyance of users we'll move to console.log. Hey Dan, Just wondering...does TwitPic have a bug or misconfiguration or is this an @anywhere bug? Thanks, -damon
Re: [twitter-dev] Re: alert() in anywhere.js
On Wed, May 19, 2010 at 11:48 AM, Damon Clinkscales sca...@pobox.com wrote: On Wed, May 19, 2010 at 1:41 PM, Dan Webb d...@twitter.com wrote: On Wed, May 19, 2010 at 11:27 AM, Steve C st...@twitpic.com wrote: Just wondering...does TwitPic have a bug or misconfiguration or is this an @anywhere bug? Javascript errors at startup time (in these cases by browser bugs in certain browsers that we don't support) are causing the initialization to terminate early leaving the client ID unset. We're going to ensure that unsupport browsers fail silently rather than triggering this alert. ETA for fix is within the hour. -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong +1 415 425 5631
[twitter-dev] Re: alert() in anywhere.js
Thanks Dan- We appreciate you your teams hard work. On May 19, 2:53 pm, Dan Webb d...@twitter.com wrote: On Wed, May 19, 2010 at 11:48 AM, Damon Clinkscales sca...@pobox.com wrote: On Wed, May 19, 2010 at 1:41 PM, Dan Webb d...@twitter.com wrote: On Wed, May 19, 2010 at 11:27 AM, Steve C st...@twitpic.com wrote: Just wondering...does TwitPic have a bug or misconfiguration or is this an @anywhere bug? Javascript errors at startup time (in these cases by browser bugs in certain browsers that we don't support) are causing the initialization to terminate early leaving the client ID unset. We're going to ensure that unsupport browsers fail silently rather than triggering this alert. ETA for fix is within the hour. -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong +1 415 425 5631
[twitter-dev] Re: alert() in anywhere.js
One of us is crazy here. If I'm not wrong console.log belongs to firebug. Which means you will get a javascript error on ALL browsers which do not have firebug installed and running. -Nischal On May 19, 11:41 pm, Dan Webb d...@twitter.com wrote: On Wed, May 19, 2010 at 11:27 AM, Steve C st...@twitpic.com wrote: We just rolled out @anywhere yesterday and some of our users are experiencing similar issues. http://twitpic.com/1p00d6 We rolled out a fix at the weekend that we fixed all the browsers that we test under but there are obviously still some browsers getting the issue. I think we'll use console.info to display these message instead of an alert. We wanted to let developers know that they needed a clientID in the most noticable way but to avoid unintended annoyance of users we'll move to console.log. Thanks, -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong +1 415 425 5631
[twitter-dev] Re: alert() in anywhere.js
I am assuming they will create a dummy function that will be used console.log is undefined- otherwise, you are very right. On May 19, 3:03 pm, nischalshetty nischalshett...@gmail.com wrote: One of us is crazy here. If I'm not wrong console.log belongs to firebug. Which means you will get a javascript error on ALL browsers which do not have firebug installed and running. -Nischal On May 19, 11:41 pm, Dan Webb d...@twitter.com wrote: On Wed, May 19, 2010 at 11:27 AM, Steve C st...@twitpic.com wrote: We just rolled out @anywhere yesterday and some of our users are experiencing similar issues. http://twitpic.com/1p00d6 We rolled out a fix at the weekend that we fixed all the browsers that we test under but there are obviously still some browsers getting the issue. I think we'll use console.info to display these message instead of an alert. We wanted to let developers know that they needed a clientID in the most noticable way but to avoid unintended annoyance of users we'll move to console.log. Thanks, -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong +1 415 425 5631
[twitter-dev] Re: alert() in anywhere.js
I think throw() seems more appropriate. Firebug and WebKit-based browsers will work with console.log(). A javascript error for an undefined function would still be better than an alert(). Larry On May 19, 12:03 pm, nischalshetty nischalshett...@gmail.com wrote: One of us is crazy here. If I'm not wrong console.log belongs to firebug. Which means you will get a javascript error on ALL browsers which do not have firebug installed and running. -Nischal On May 19, 11:41 pm, Dan Webb d...@twitter.com wrote: On Wed, May 19, 2010 at 11:27 AM, Steve C st...@twitpic.com wrote: We just rolled out @anywhere yesterday and some of our users are experiencing similar issues. http://twitpic.com/1p00d6 We rolled out a fix at the weekend that we fixed all the browsers that we test under but there are obviously still some browsers getting the issue. I think we'll use console.info to display these message instead of an alert. We wanted to let developers know that they needed a clientID in the most noticable way but to avoid unintended annoyance of users we'll move to console.log. Thanks, -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong +1 415 425 5631
[twitter-dev] Re: alert() in anywhere.js
Phew... I would have agreed if u guys had said I was crazy too... I follow these threads to understand and learn new stuff... :) -Nischal On May 20, 12:07 am, Steve C st...@twitpic.com wrote: I am assuming they will create a dummy function that will be used console.log is undefined- otherwise, you are very right. On May 19, 3:03 pm, nischalshetty nischalshett...@gmail.com wrote: One of us is crazy here. If I'm not wrong console.log belongs to firebug. Which means you will get a javascript error on ALL browsers which do not have firebug installed and running. -Nischal On May 19, 11:41 pm, Dan Webb d...@twitter.com wrote: On Wed, May 19, 2010 at 11:27 AM, Steve C st...@twitpic.com wrote: We just rolled out @anywhere yesterday and some of our users are experiencing similar issues. http://twitpic.com/1p00d6 We rolled out a fix at the weekend that we fixed all the browsers that we test under but there are obviously still some browsers getting the issue. I think we'll use console.info to display these message instead of an alert. We wanted to let developers know that they needed a clientID in the most noticable way but to avoid unintended annoyance of users we'll move to console.log. Thanks, -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong +1 415 425 5631
[twitter-dev] Re: alert() in anywhere.js
Any update on this? I'm getting it on a site for certain users/ browsers even though it's all configured correctly as well. The alerts are very intrusive for a production website when things are configured correctly. Especially with user emails rolling in complaining about getting the error. On May 16, 10:22 am, Dan Webb d...@twitter.com wrote: This does sound like a regression of some kind. We'll get this fixed ASAP. On Sat, May 15, 2010 at 3:41 PM, JohnB johnfakor...@yahoo.com wrote: Are we really talking about incorrect installations here? Twitter's own @Anywhere documentation page (http://dev.twitter.com/anywhere/ begin) is throwing this same error in older browsers, including Chrome 3.0.195. -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong+1 415 425 5631
[twitter-dev] Re: alert() in anywhere.js
Are we really talking about incorrect installations here? Twitter's own @Anywhere documentation page (http://dev.twitter.com/anywhere/ begin) is throwing this same error in older browsers, including Chrome 3.0.195. On May 15, 6:26 pm, Abraham Williams 4bra...@gmail.com wrote: I agree that @Anywhere should degrade gracefully when configured properly on unsupported platforms and not prompt incorrect alert()s. But I do think alert()s are probably the best way to notify developers of incorrect installations. Abraham On Sat, May 15, 2010 at 11:55, Larry la...@topsy.com wrote: I can reliably reproduce this with Firefox 3.0.8 at the following url: http://cornsyrup.org/~larry/anywhere/index.html Error console is reporting S.get is not a function Larry On May 15, 11:31 am, Larry la...@topsy.com wrote: Our site has been running @anywhere for over a week now without error. Yesterday my coworker was getting the alert(). He is running an older version of Firefox (3.0.8) on Ubuntu, so there might be another cause other than missing clientID or version? I still believe alert() is intrusive, especially for this case where it works fine except for this edge case. Instead of users complaining about broken hovercards, they are complaining about alert dialogs. Larry On May 14, 8:38 pm, Abraham Williams 4bra...@gmail.com wrote: Both of which are issues that will pretty much stop @Anywhere from working and need to be noticed as soon as possible at installation. Hiding them in console.log will make it more likely that @Anywhere will be installe improperly and the admins will only find out when users complain. Abraham On Fri, May 14, 2010 at 15:57, Larry la...@topsy.com wrote: I just came across a coworker's browser that triggered an alert() call from anywhere.js. While okay for development, the use of alert() is not friendly for production websites. Could these be converted console.log() or some other benign mechanism? Grepping through anywhere.js I found two instances of alert(): alert(To set up @anywhere, please provide a client ID); alert(No version matching +Z); Cheers Larry -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private. -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private.
Re: [twitter-dev] Re: alert() in anywhere.js
This does sound like a regression of some kind. We'll get this fixed ASAP. On Sat, May 15, 2010 at 3:41 PM, JohnB johnfakor...@yahoo.com wrote: Are we really talking about incorrect installations here? Twitter's own @Anywhere documentation page (http://dev.twitter.com/anywhere/ begin) is throwing this same error in older browsers, including Chrome 3.0.195. -- Dan Webb Front-end Engineer, Platform d...@twitter.com / @danwrong +1 415 425 5631
[twitter-dev] Re: alert() in anywhere.js
Our site has been running @anywhere for over a week now without error. Yesterday my coworker was getting the alert(). He is running an older version of Firefox (3.0.8) on Ubuntu, so there might be another cause other than missing clientID or version? I still believe alert() is intrusive, especially for this case where it works fine except for this edge case. Instead of users complaining about broken hovercards, they are complaining about alert dialogs. Larry On May 14, 8:38 pm, Abraham Williams 4bra...@gmail.com wrote: Both of which are issues that will pretty much stop @Anywhere from working and need to be noticed as soon as possible at installation. Hiding them in console.log will make it more likely that @Anywhere will be installe improperly and the admins will only find out when users complain. Abraham On Fri, May 14, 2010 at 15:57, Larry la...@topsy.com wrote: I just came across a coworker's browser that triggered an alert() call from anywhere.js. While okay for development, the use of alert() is not friendly for production websites. Could these be converted console.log() or some other benign mechanism? Grepping through anywhere.js I found two instances of alert(): alert(To set up @anywhere, please provide a client ID); alert(No version matching +Z); Cheers Larry -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private.
[twitter-dev] Re: alert() in anywhere.js
I can reliably reproduce this with Firefox 3.0.8 at the following url: http://cornsyrup.org/~larry/anywhere/index.html Error console is reporting S.get is not a function Larry On May 15, 11:31 am, Larry la...@topsy.com wrote: Our site has been running @anywhere for over a week now without error. Yesterday my coworker was getting the alert(). He is running an older version of Firefox (3.0.8) on Ubuntu, so there might be another cause other than missing clientID or version? I still believe alert() is intrusive, especially for this case where it works fine except for this edge case. Instead of users complaining about broken hovercards, they are complaining about alert dialogs. Larry On May 14, 8:38 pm, Abraham Williams 4bra...@gmail.com wrote: Both of which are issues that will pretty much stop @Anywhere from working and need to be noticed as soon as possible at installation. Hiding them in console.log will make it more likely that @Anywhere will be installe improperly and the admins will only find out when users complain. Abraham On Fri, May 14, 2010 at 15:57, Larry la...@topsy.com wrote: I just came across a coworker's browser that triggered an alert() call from anywhere.js. While okay for development, the use of alert() is not friendly for production websites. Could these be converted console.log() or some other benign mechanism? Grepping through anywhere.js I found two instances of alert(): alert(To set up @anywhere, please provide a client ID); alert(No version matching +Z); Cheers Larry -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private.
Re: [twitter-dev] Re: alert() in anywhere.js
I agree that @Anywhere should degrade gracefully when configured properly on unsupported platforms and not prompt incorrect alert()s. But I do think alert()s are probably the best way to notify developers of incorrect installations. Abraham On Sat, May 15, 2010 at 11:55, Larry la...@topsy.com wrote: I can reliably reproduce this with Firefox 3.0.8 at the following url: http://cornsyrup.org/~larry/anywhere/index.html Error console is reporting S.get is not a function Larry On May 15, 11:31 am, Larry la...@topsy.com wrote: Our site has been running @anywhere for over a week now without error. Yesterday my coworker was getting the alert(). He is running an older version of Firefox (3.0.8) on Ubuntu, so there might be another cause other than missing clientID or version? I still believe alert() is intrusive, especially for this case where it works fine except for this edge case. Instead of users complaining about broken hovercards, they are complaining about alert dialogs. Larry On May 14, 8:38 pm, Abraham Williams 4bra...@gmail.com wrote: Both of which are issues that will pretty much stop @Anywhere from working and need to be noticed as soon as possible at installation. Hiding them in console.log will make it more likely that @Anywhere will be installe improperly and the admins will only find out when users complain. Abraham On Fri, May 14, 2010 at 15:57, Larry la...@topsy.com wrote: I just came across a coworker's browser that triggered an alert() call from anywhere.js. While okay for development, the use of alert() is not friendly for production websites. Could these be converted console.log() or some other benign mechanism? Grepping through anywhere.js I found two instances of alert(): alert(To set up @anywhere, please provide a client ID); alert(No version matching +Z); Cheers Larry -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private. -- Abraham Williams | Developer for hire | http://abrah.am @abraham | http://projects.abrah.am | http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private.
[twitter-dev] Re: alert() in anywhere.js
Firefox 3.X is a supported browser for @anywhere and my example is properly configured, yet it triggered when it wasn't supposed to. This highlights my point of why alert() not a good choice for notification of incorrect installations. Instead maybe it should use throw(). That would be more useful to a developer and not intrusive to a user. Larry On May 15, 3:26 pm, Abraham Williams 4bra...@gmail.com wrote: I agree that @Anywhere should degrade gracefully when configured properly on unsupported platforms and not prompt incorrect alert()s. But I do think alert()s are probably the best way to notify developers of incorrect installations. Abraham On Sat, May 15, 2010 at 11:55, Larry la...@topsy.com wrote: I can reliably reproduce this with Firefox 3.0.8 at the following url: http://cornsyrup.org/~larry/anywhere/index.html Error console is reporting S.get is not a function Larry On May 15, 11:31 am, Larry la...@topsy.com wrote: Our site has been running @anywhere for over a week now without error. Yesterday my coworker was getting the alert(). He is running an older version of Firefox (3.0.8) on Ubuntu, so there might be another cause other than missing clientID or version? I still believe alert() is intrusive, especially for this case where it works fine except for this edge case. Instead of users complaining about broken hovercards, they are complaining about alert dialogs. Larry On May 14, 8:38 pm, Abraham Williams 4bra...@gmail.com wrote: Both of which are issues that will pretty much stop @Anywhere from working and need to be noticed as soon as possible at installation. Hiding them in console.log will make it more likely that @Anywhere will be installe improperly and the admins will only find out when users complain. Abraham On Fri, May 14, 2010 at 15:57, Larry la...@topsy.com wrote: I just came across a coworker's browser that triggered an alert() call from anywhere.js. While okay for development, the use of alert() is not friendly for production websites. Could these be converted console.log() or some other benign mechanism? Grepping through anywhere.js I found two instances of alert(): alert(To set up @anywhere, please provide a client ID); alert(No version matching +Z); Cheers Larry -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private. -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private.
[twitter-dev] Re: alert() in anywhere.js
Isn't console.log() specific to firebug? #JustSaying :) On May 16, 4:43 am, Larry la...@topsy.com wrote: Firefox 3.X is a supported browser for @anywhere and my example is properly configured, yet it triggered when it wasn't supposed to. This highlights my point of why alert() not a good choice for notification of incorrect installations. Instead maybe it should use throw(). That would be more useful to a developer and not intrusive to a user. Larry On May 15, 3:26 pm, Abraham Williams 4bra...@gmail.com wrote: I agree that @Anywhere should degrade gracefully when configured properly on unsupported platforms and not prompt incorrect alert()s. But I do think alert()s are probably the best way to notify developers of incorrect installations. Abraham On Sat, May 15, 2010 at 11:55, Larry la...@topsy.com wrote: I can reliably reproduce this with Firefox 3.0.8 at the following url: http://cornsyrup.org/~larry/anywhere/index.html Error console is reporting S.get is not a function Larry On May 15, 11:31 am, Larry la...@topsy.com wrote: Our site has been running @anywhere for over a week now without error. Yesterday my coworker was getting the alert(). He is running an older version of Firefox (3.0.8) on Ubuntu, so there might be another cause other than missing clientID or version? I still believe alert() is intrusive, especially for this case where it works fine except for this edge case. Instead of users complaining about broken hovercards, they are complaining about alert dialogs. Larry On May 14, 8:38 pm, Abraham Williams 4bra...@gmail.com wrote: Both of which are issues that will pretty much stop @Anywhere from working and need to be noticed as soon as possible at installation. Hiding them in console.log will make it more likely that @Anywhere will be installe improperly and the admins will only find out when users complain. Abraham On Fri, May 14, 2010 at 15:57, Larry la...@topsy.com wrote: I just came across a coworker's browser that triggered an alert() call from anywhere.js. While okay for development, the use of alert() is not friendly for production websites. Could these be converted console.log() or some other benign mechanism? Grepping through anywhere.js I found two instances of alert(): alert(To set up @anywhere, please provide a client ID); alert(No version matching +Z); Cheers Larry -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private. -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private.
Re: [twitter-dev] Re: alert() in anywhere.js
Chrome has it too. On Sun, May 16, 2010 at 03:20, nischalshetty nischalshett...@gmail.comwrote: Isn't console.log() specific to firebug? #JustSaying :) On May 16, 4:43 am, Larry la...@topsy.com wrote: Firefox 3.X is a supported browser for @anywhere and my example is properly configured, yet it triggered when it wasn't supposed to. This highlights my point of why alert() not a good choice for notification of incorrect installations. Instead maybe it should use throw(). That would be more useful to a developer and not intrusive to a user. Larry On May 15, 3:26 pm, Abraham Williams 4bra...@gmail.com wrote: I agree that @Anywhere should degrade gracefully when configured properly on unsupported platforms and not prompt incorrect alert()s. But I do think alert()s are probably the best way to notify developers of incorrect installations. Abraham On Sat, May 15, 2010 at 11:55, Larry la...@topsy.com wrote: I can reliably reproduce this with Firefox 3.0.8 at the following url: http://cornsyrup.org/~larry/anywhere/index.html Error console is reporting S.get is not a function Larry On May 15, 11:31 am, Larry la...@topsy.com wrote: Our site has been running @anywhere for over a week now without error. Yesterday my coworker was getting the alert(). He is running an older version of Firefox (3.0.8) on Ubuntu, so there might be another cause other than missing clientID or version? I still believe alert() is intrusive, especially for this case where it works fine except for this edge case. Instead of users complaining about broken hovercards, they are complaining about alert dialogs. Larry On May 14, 8:38 pm, Abraham Williams 4bra...@gmail.com wrote: Both of which are issues that will pretty much stop @Anywhere from working and need to be noticed as soon as possible at installation. Hiding them in console.log will make it more likely that @Anywhere will be installe improperly and the admins will only find out when users complain. Abraham On Fri, May 14, 2010 at 15:57, Larry la...@topsy.com wrote: I just came across a coworker's browser that triggered an alert() call from anywhere.js. While okay for development, the use of alert() is not friendly for production websites. Could these be converted console.log() or some other benign mechanism? Grepping through anywhere.js I found two instances of alert(): alert(To set up @anywhere, please provide a client ID); alert(No version matching +Z); Cheers Larry -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private. -- Abraham Williams | Developer for hire |http://abrah.am @abraham |http://projects.abrah.am|http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private. -- Abraham Williams | Developer for hire | http://abrah.am @abraham | http://projects.abrah.am | http://blog.abrah.am This email is: [ ] shareable [x] ask first [ ] private.
