Got ratty with the automated connectivity check / captive portal 'helper' today whilst bouncing my internet.

Found some instructions at fedoraforum.org (which themselves link to blogs.gnome.org ...

Key fix applied: modify "/etc/NetworkManager/NetworkManager.conf" and add this at the bottom:

[connectivity]
uri=
interval=0

I trust that I won't see it again. :|

Down the rabbit hole I went today, and tripped over this page at RSM which discusses the issues importing electrical, electronic and radio devices into NZ from a Government perspective. I noted "suppliers of radio transmitting products must hold a Licence to Supply.". Makes me wonder whether Trademe and such are watching for this sort of business, what with people routinely purchasing transmission kit to resell it...

For the first time, I've recently had cause to parse a large number of log files, looking for (and removing) fields and shipping the output into dynamically-named additional files.

The first challenge was to reduce the log files down to just the ones that contained relevant text. Usually i'd have done this with a simple 'grep' (perhaps then piped to another grep to further narrow).

I discovered that there's a tool 'somewhat' similar to grep, in the 'cmd' suite called 'findstr' and that I was able to run search multiple files by using a 'for' loop:

> for /F "tokens=*" %A in (listoflogfiles.txt) do findstr /s %A * >>output_%A.txt
(to simplify my life I create a text file which contained a line-by-line list of all the files I wanted to parse (dir /w >>listoflogfiles.txt) which allowed me to manually tweak the list)

Of interest: tokens is the number of variables on each line to deal with. So '*' means all of them. There's an additional parameter called 'delims' that can be used to specify the delimiter... so this actually has some similarities to bash cut.

Also of interest: findstr /s means it's recursive (search all subdirs)

So anyway. This gave me all the data in each line matching the string, but I wanted to limit the output fields to just ones useful in my application. The way I've done this historically is with a simple 'while line' loop in bash, together with something like 'cut -d " " -f 2,3,6-' >>$line.txt (using a blank space as a delimiter, show me only fields 2, 3, and all fields after and including 6 based on that delimiter, output to a file named based on the input value of the line).

I often see people asking how to change their email address for mailing lists.
Most mailing lists - Mailman notably - tends to treat each email address as a given subscription. So it's actually entirely possible to treat the procedure as a two-parter, being 'subscribing with a new address' followed by 'unsubscribing your old address'. More modern installations also support 'changing' your email address... so see Alternative: below.

Two-part process.
This is useful if you want to ensure delivery to the new address before cancelling the old one (a 'soft' transition).

Phase 1: Subscribe your new address

Step 1: Visit the mailing list information page - usually in the footer of every email to the list, is a link.

-----
This is my Mailing list
mailinglist@email.example.com
http://mailinglist.example.com/mailman/listinfo/mailinglist

If you click on the hyperlink you'll see a page about the list. On the page will be links to visit the Archive (may require you to authenticate with a password), instructions on how to post to the list (what email address to send to), and a form through which you can subscribe by entering your name and email address.

Step 2: Fill in the "Subscribing to MailingList" form with your name and email address. If you don't also submit a password, a random one will be generated and emailed back to you with your confirmation of subscription.

Step 3: Check your email for a message asking you to confirm that the address is infact yours. It'll have a link and/or a text string that you should use to confirm your subscription. (This is to prevent people subscribing you without your consent). Once you've done this, your new email address will start receiving traffic.... note that for some mailing lists, a Moderator or Administrator may need to further approve your subscription.

Phase 2: Remove the old address (aka 'how to unsubscribe')

Step 1: Visit the mailing list information page (as above).

Surprised I didn't put this out there earlier...

Freeparking (a large domain name registrar in New Zealand) have been using Cyber Security for marketing purposes, pushing the .kiwi tld.

Today I got the following within (yet another) marketing plug from them:

"Cyber threats are on the rise in New Zealand and many small to medium business are leaving themselves vulnerable by not securing the core domains recognised in New Zealand. At Freeparking, we want to help you protect your brand which is why we had these domains put on hold for you - but time is running out, your .KIWI domains are expiring soon, so ensure you claim and renew your domains now!"

What they actually did was pre-emptively register the .kiwi variations of domain names registered by their customers, a year ago, and the annual renewal is coming due.

I finally put my finger on the change to vertical scrollbars in Ubuntu that I wasn't expecting.

It's described at this stackexchange post and also on a Mozilla page. I'm giving the settings.ini option a go.

Not that I don't think the 'warp' idea is a bad one. But they should've put it on the right click by default, I feel.. not the left.

So.. this is proving a bit of a peeve... advice appreciated:

- HP Laptop set up with BIOS Clock showing local time.
- Ubuntu Linux set up (17.10) and shows local time correctly.
- Windows 10 (OEM, albeit with less HDD assigned) keeps putting itself back in time by 13 hours! (We're in DST, so we're UTC +13. Coincidence?)

If I toggle Windows 10 auto-sync off-and-on again and it resync's via NTP, the clock goes back to normal.
I can't make sense of it. And Windows 10 seem to have stunted the ability to manually get the clock to sync.

I spend about half my time in Windows and the other half in Linux, so this is bloody annoying.

Editors Note: This page originally cited that Reddit was the source of the popup ads. Having closed Reddit down for a while it happened again - see amendment - and the only common factor was Memory Alpha (The Star Trek Wiki) - which rings true as I had suspected similar some time ago. I've been re-watching Star Trek via Netflix and following the production notes on Memory-Alpha as I go, so there's been a tab floating on one of their content pages for a while. :(

Mobile browser has been popping up with fake ads proporting to be flogging large-value Countdown supermarket vouchers (but only if you fill it in real fast, 2 minutes!)...

The content is hosted on cloudfront (AWS) and when I later reopened the URL to investigate it I got redirected to www.kiwiprizes.com - a domain hosted by Cloudflare, and apparently registered to:

Domain Name: kiwiprizes.com
Registry Domain ID: 1941599927_DOMAIN_COM-VRSN
Registrar WHOIS Server: whois.psi-usa.info
Registrar URL: http://www.psi-usa.info
Updated Date: 2017-08-13T14:21:02Z
Creation Date: 2015-06-24T13:02:50Z
Registrar Registration Expiration Date: 2018-06-24T13:02:50Z
Registrar: PSI-USA, Inc. dba Domain Robot
Registrar IANA ID: 151
Registrar Abuse Contact Email: domain-abuse@psi-usa.info
Registrar Abuse Contact Phone: +49.94159559482

Registrant Name: Jimmy Nguyen
Registrant Organization: AP Marketing Asia Pacific Pte Ltd
Registrant Street: Unit 706 7/F South Sea Centre Tower 2, 75 Mody Road
Registrant City: Hong Kong
Registrant State/Province: Kowloon
Registrant Postal Code: 852
Registrant Country: HK
Registrant Phone: +49.615185080
Registrant Phone Ext:
Registrant Fax: +49.61518508111
Registrant Fax Ext:
Registrant Email: domains@spark5.de

Registry Tech ID:
Tech Name: Department Systemadministration

Had a bit of a tinker with my daughters Chromebook today.
In particular wanted to see if I could make it boot from USB (to find a way of running an OS on it other than ChromeOS, and without actually hosing the ChromeOS instance in the process.

Managed to generally follow the instructions I found at Chromium.org to enable Developer mode.

Once done, you boot it up as usual (Ctrl-D to keep booting) and once started, tied to Wifi and logged in with your Google account, you can use Ctrl-Alt-T to run a ChromeOS Terminal, and from there type 'shell' to get a bash shell.
The trap is that you're then logged in as 'chronos' and to be able to act as root you need to be able to su upward... requiring the knowledge of the password for chronos. The instructions I then found on Reddit (Ctrl-Alt-RightArrowFunction) lets you open a rootshell using the previously set root password, and then set a password for the user chronos, which can then be used to fully interact without tasking out of the Window manager...

Unfortunately the instructions to let the thing boot of USB don't appear to work, and it appears, i'm not the only one to confront this. .

That's about as far as i've gotten, I don't particularly want to brick it, nor have to factory default it as such. Just be nice to have an alternative boot option occasionally. Sigh...

Today's fail was brought to you by Microsoft.

Above from https://twitter.com/vanbroup/status/869123089820516352. - Click on the pic.twitter.com link to see the screenshot that makes it all very obvious.

See also

https://developer.microsoft.com/en-us/microsoft-edge/platform/issues/12163379/

It appears only Firefox and Chrome appear to even validate this, but it means that numerous Microsoft systems including Office 365 (Sharepoint Online, Onedrive for Business, Azure and others) simply aren't letting Firefox users in (Because Firefox treats the error as Fatal). Apparently Chrome works-but-regards-as-insecure and IE still works without showing the user anything.

Unfortunately the workaround is to set the security.ssl.enable_ocsp_stapling. to false in about:config in Firefox (disabling the check).

https://answers.microsoft.com/en-us/outlook_com/forum/osecurity-oother/error-when-trying-to-log-in-to-outlook-using/9f8dd00f-e4d8-4984-ac17-ecf85cacbac6

I suppose the trick is getting all those users who implement the workaround to turn the check back on when Microsoft fix it?

Even one of the biggest technology companies in the world can fail at the basics.

Syndicate content