[webif] Web Interface 1.4.x

[MymsMan]

I've just uploaded webif beta 1.4.9-1, containing a number of fixes and improvements from @/df, @prpr and @MymsMan

A list of changes can be viewed at:

webif

webif

git.hpkg.tv

Thanks to the contributors, apologies for the backlog!

You'll need a long rest after a marathon git update session!

 

[everthewatcher]

Look like we have problem, or at least I have. Just updated to webif beta 1.4.9-1 and now on checking for updates the Package Management Results box stalls at:

Processing request...

[Edit] Rebooted after a full power down (heard HD stop) just in case and no difference.

[Edit 2] It doesn't stall but takes minutes instead of seconds to complete.

 

[Black Hole]

You're right, it does seem slow - but that could be at the server end.

 

[everthewatcher]

You're right, it does seem slow - but that could be at the server end.

The check for updates immediately before the update took the usual couple of seconds. Just timed it and it's now 190 seconds.

 

[4ndy]

Mine’s been going 15 mins. It could be everyone trying it at the same time.

 

[MymsMan]

Any messages in webif-error.log?

Yes, Timesouts

8    at file "/mod/webif/lib/pkg.class", line 228
7    in procedure 'pkg' called at file "/mod/webif/lib/pkg.class", line 248
6    in procedure 'pkg' called at file "/mod/webif/cgi-bin/opkg.jim", line 73
5    /mod/webif/lib/pkg.class:228: Error: hpkg.tv:http: connect: Connection timed out
4    at file "/mod/webif/lib/pkg.class", line 228
3    in procedure 'pkg' called at file "/mod/webif/lib/pkg.class", line 248
2    in procedure 'pkg' called at file "/mod/webif/cgi-bin/opkg.jim", line 73
1    /mod/webif/lib/pkg.class:228: Error: hpkg.tv:http: connect: Connection timed out

 

[prpr]

Mine are virtually instantaneous, as is opkg update from the command line.

 

[everthewatcher]

Yes, Timesouts

Same as here. And I'm off to bed.

 

[prpr]

Any messages in webif-error.log?

1 /mod/webif/lib/pkg.class:228: Error: hpkg.tv:http: connect: Connection timed out[/CODE]

Interesting.
Can you try this and see how it compares to mine:

humax ~ # jimsh
Welcome to Jim version 0.80
. socket stream hpkg.tv:80
Not a valid address: hpkg.tv:80
[error] . socket stream hpkg.tv:http
::aio.sock4
.

 

[Black Hole]

For some reason, mine is Jim version 0.79:

HDRFOX1# jimsh                                                                                     
Welcome to Jim version 0.79                                                                       
. socket stream hpkg.tv:80
::aio.sock4                                                                                       
. socket stream hpkg.tv:http

...and then it just froze.

 

[prpr]

Yes, it times out eventually, as the previous posts show.
Here's how to fix it:

humax# sed -i "s/:http/:80/" /mod/webif/lib/pkg.class

Sorry. That one's my fault.

 

[Black Hole]

Why am I showing a different Jim version?

 

[everthewatcher]

Here's how to fix it...

A bit of hand-holding here please - what does one do with that and what effect does it have?
[Later] Ah, I've looked up SED. And to think that I sometimes use 's/this/that' in a comment to correct an uneditable posting.

 

[Black Hole]

what does one do with that and what effect does it have?

Enter sed -i "s/:http/:80/" /mod/webif/lib/pkg.class on the command line, obtained via Telnet or (with webshell installed) WebIf >> Diagnostics >> Command Line.

I've got four machines to do - I'll wait for the movie to come out... or is there a Catch 22 here - has this crippled updates?

 

[everthewatcher]

Enter...

Yes, I figured that out once I'd woken up. I'm not a Linux person but the 's/this/that' construction has been used on Cix for as long as I can remember to correct a posting.

I've got four machines to do - I'll wait for the movie to come out... or is there a Catch 22 here - has this crippled updates?

Funny thing is I had an 'Is this a good idea?' moment before updating last night. Crippling updates would be a bad thing.

 

[Black Hole]

Funny thing is I had an 'Is this a good idea?' moment before updating last night.

The whole point of being beta testers is to catch things like this. We have to be willing to take risks, and report what we find (even if some of us poke it less than others). WebIF 1.4.9-1 is only available to those signed up as beta package testers (by installing opkg-beta - see https://hummy.tv/forum/threads/beta-participation.7785/).

 

[everthewatcher]

The whole point of being beta testers is to catch things like this.

Yes, I realise this - I used to be a beta tester on some OLR software many years back.

 

[prpr]

Why am I showing a different Jim version?

Because I'm running a new build. Something is different between 0.79 and 0.80 (or the builds thereof) which has upset this. I can't see what it is looking at the source code, but 0.79 definitely seems broken somehow to me as using the :http version should generate an error earlier in the process - it doesn't and tries to connect to port 0 on the web server anyway, which is what eventually causes the timeout error.
All this should really be in a thread in the Beta section.

I've got four machines to do

It's copy once and paste four times. Do them all in less than a minute. Even easier if you use ssh (from a Linux or W10 machine) rather than telnet as you can put it all on the command line (you need to have set up dropbear properly first though, but well worth doing) and repeat for each machine.

has this crippled updates?

I don't think so. Even so, you could do opkg update from the command line, followed by the GUI Upgrades tab, avoiding the "Check for updates" button. But if you were doing the former, you might as well do opkg upgrade whilst you were there.

 

[Black Hole]

It's copy once and paste four times. Do them all in less than a minute.

Yes, no doubt, but somebody needs to test the roll-out which corrects the bug. If it's no major inconvenience, there's no emergency.

 

[Brian]

Here's how to fix it:

humax# sed -i "s/:http/:80/" /mod/webif/lib/pkg.class

Sorry. That one's my fault.

I have used this command to fix both boxes that had the issue.