x/tools/gopls: report better error messages for invalid flags

[frankful]

gopls version: v0.5.1
gopls flags: -remote=aut -logfile=auto -debug=:0 -rpc.trace

ATTENTION: PLEASE PROVIDE THE DETAILS REQUESTED BELOW.

Describe what you observed.

Please attach the stack trace from the crash.
A window with the error message should have popped up in the lower half of your screen.
Please copy the stack trace and error messages from that window and paste it in this issue.

OPTIONAL: If you would like to share more information, you can attach your complete gopls logs.

NOTE: THESE MAY CONTAIN SENSITIVE INFORMATION ABOUT YOUR CODEBASE.
DO NOT SHARE LOGS IF YOU ARE WORKING IN A PRIVATE REPOSITORY.

<OPTIONAL: ATTACH LOGS HERE>

[stamblerre]

Edit: I'm sorry I misread the version here.
I'm closing this, as the issue is empty.

@frankful: Please reopen if you can share the stack trace of the gopls crash (found by navigating to View -> Output -> Tasks -> gopls (server)).

[hyangah]

@frankful Please check the error message printed out in Output -> gopls (server) channel.
It looks like the value you pass for -remote flag is invalid.

@stamblerre Crash when the -remote flag value is invalid, is not very visible (just one line :-). Can we make it more visually standing out?

[stamblerre]

@stamblerre Crash when the -remote flag value is invalid, is not very visible (just one line :-). Can we make it more visually standing out?

Sure, let me transfer this to the gopls issue tracker.

[stamblerre]

@hyangah: I guess the real issue is that the crash is not a panic, so it doesn't get picked up by your logic to extract stack traces. The current error message (gopls: forwarder: connecting to remote: dialing remote: dial tcp: address aut: missing port in address) seems pretty good to me, it's just that it doesn't get automatically attached to the issue.

/cc @findleyr

[hyangah]

@stamblerre I don't worry too much about the issue tracker contents (we already include the gopls flag values in the issue so the setting error was pretty obvious to me :-)

Ideally I want users to figure out this type of setting mistakes by themselves even before sending an issue report to us. I want to know if there is a better way to display this error message. Currently, the error message is buried in stderr and it's not immediately obvious this is significant enough to crash the server. Can it send 'Error' -level logs? Or prefix the message with 'FATAL', or something that sounds serious?

gopls: forwarder: connecting to remote: dialing remote: dial tcp 127.0.0.1:12345: connect: connection refused
[Info  - 6:12:39 PM] Connection to server got closed. Server will restart.
gopls: forwarder: connecting to remote: dialing remote: dial tcp 127.0.0.1:12345: connect: connection refused
[Info  - 6:12:43 PM] Connection to server got closed. Server will restart.
gopls: forwarder: connecting to remote: dialing remote: dial tcp 127.0.0.1:12345: connect: connection refused
[Info  - 6:12:47 PM] Connection to server got closed. Server will restart.
gopls: forwarder: connecting to remote: dialing remote: dial tcp 127.0.0.1:12345: connect: connection refused
[Info  - 6:12:51 PM] Connection to server got closed. Server will restart.
gopls: forwarder: connecting to remote: dialing remote: dial tcp 127.0.0.1:12345: connect: connection refused
[Error - 6:12:55 PM] Connection to server got closed. Server will not be restarted.