Closed budde377 closed 7 years ago
Thanks for catching this :) Can you update the tests to reflect this change?
Will do!
Here you go.
thanks :)
ok published in version 1.5.0 (possible breaking change if people are still using old binaries)
Nice
On Mon, Oct 3, 2016, 14:11 Peter Halliday notifications@github.com wrote:
ok published in version 1.5.0 (possible breaking change if people are still using old binaries)
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/pghalliday/node-BrowserStackTunnel/pull/27#issuecomment-251090686, or mute the thread https://github.com/notifications/unsubscribe-auth/ADVCy8e4Dy8vEAaTktPnsP-LPwmUQknDks5qwPD6gaJpZM4KMXQX .
This just broke our deploy pipeline and took time to nail down. I strongly encourage you to follow NPM community best practices around semantic versioning, specifically always publishing breaking changes as major version bumps. Thank you.
@serhalp How did it fail? Was it because of old binaries?
@serhalp ah yes, i forgot I was at 1.x and not 0.x - my bad
@serhalp - i can delete version 1.5.0 and release 2.0.0. I think that should fix your pipeline (and stop you picking up 1.5.0)
@serhalp - ok sorry about that, now released as 2.0.0
Here's the error we were seeing (makes this PR findable by search engines):
BrowserStackLocal v6.1
*** Error: Could not connect to --only!
Configuration Options:
-v
Provides verbose logging
-f
If you want to test local folder rather internal server
-h
Prints this help
-version
Displays the version
-force
Kill other running Browserstack Local
-only
Restricts Local Testing access to specified local servers and/or folders
-forcelocal
Route all traffic via local machine
-onlyAutomate
Disable Live Testing and Screenshots, just test Automate
-proxyHost HOST
Hostname/IP of proxy, remaining proxy options are ignored if this option is absent
-proxyPort PORT
Port for the proxy, defaults to 3128 when -proxyHost is used
-proxyUser USERNAME
Username for connecting to proxy (Basic Auth Only)
-proxyPass PASSWORD
Password for USERNAME, will be ignored if USERNAME is empty or not specified
-localIdentifier SOME_STRING
If doing simultaneous multiple local testing connections, set this uniquely for different processes
To test an internal server, run:
./BrowserStackLocal <KEY>
Example:
./BrowserStackLocal DsVSdoJPBi2z44sbGFx1
To test HTML files, run:
./BrowserStackLocal -f <KEY> <full path to local folder>
Example:
./BrowserStackLocal -f DsVSdoJPBi2z44sbGFx1 /Applications/MAMP/htdocs/example/
View more configuration options at https://www.browserstack.com/local-testing
at BrowserStackTunnel.exit (node_modules/browserstacktunnel-wrapper/src/BrowserStackTunnel.js:136:28)
at Process.ChildProcess._handle.onexit (internal/child_process.js:200:12)
I'm not sure how I feel about having unpublished that version... You may have just broken someone else's builds, but at the expense of avoiding breaking countless others? Unclear. It may have been cleaner to npm deprecate
1.5.0 with a message along the lines of "accidentally published breaking change in minor bump" – much easier to debug than a spontaneous 404 from npm. Oh well.
@serhalp - ah well it was only up for 8 hours. I'll look into npm deprecate for the future though
Notice, the newest binaries are version 6.2
@pghalliday I tried downloading the binaries again and now I'm getting 6.1. It might just be that BrowserStack published the binaries by mistake. This is a mess.
I have written to BrowserStack asking about this.
oh dear
Okay.. So I got a response from them saying:
Yes, we released a new version of the binary (v6.2) yesterday. Unfortunately, this resulted in Karma tests failing due to the way Karma launches the Local Testing binary. We reverted to the pervious version (v6.1) immediately. Apologies for the inconvenience caused.
They promised to keep me posted, so I'll stay on this and check-in when I hear from them again.
ok well i don't expect any more updates before they figure that out, so I think we're safe due to the major version update.
Interestingly, I think Karma uses this library so I hope the problem is not here :s
Great! It was exactly failing tests in karma that made me create the initial pull request. But I have ensured that they are using v. 1.4.* until this is fixed.
On Tue, Oct 4, 2016 at 2:59 PM Peter Halliday notifications@github.com wrote:
ok well i don't expect any more updates before they figure that out, so I think we're safe due to the major version update.
Interestingly, I think Karma uses this library so I hope the problem is not here :s
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/pghalliday/node-BrowserStackTunnel/pull/27#issuecomment-251380608, or mute the thread https://github.com/notifications/unsubscribe-auth/ADVCy4mJiN2oGmB0nBHuCcSokvJ0w4TXks5qwk2ngaJpZM4KMXQX .
ok cool
I got a response from BrowserStack. They are promising that backwards-compatibility will be honoured for the upcoming release. Thus the new binaries should work with version 1.4 and 2.0! I'll expect them to write again when the new binaries are released and then I'll write here again.
thanks!
From BrowserStack
Just an update - we have released a new version of the binary v6.2. As shared earlier, this has backward compatibility and the old commands/parameters should work.
I guess everything is alright now and people can use v1.4 as v2.
Awesome, many thanks for coordinating this :)
When running browser-stack local I get the following errr:
This seems to be because of the arguments of the binary being updated:
This pull request should update to new args, assuming that
hosts
are equivalent to theBrowserstackLocal --only
argument