Seems I'm the idiot then, missed the "gateway" bit... so this basically sits between something providing VNC/RDP and makes it accessible with just a web server?

It does say... "once Guacamole is installed on a server" so one would guess you do need Guacamole installed on the controlled machine...

One would hope Tesco would cover any costs incurred by customers that are a direct result of this. I know banks in the past have cancelled non-arranged overdraft fees when they've screwed up, but that's easier to do as it's cancelling a charge rather than actually paying money out.

One would hope, but again time is a factor. I've had a weird "I'll be OK soon but for now I'm in trouble" situation before with Bank of Scotland back when I was a student. BOS randomly decided overnight that instead of having an agreed-upon student overdraft limit of GBP 2000 I had none whatsoever and needed to start paying them the full balance immediately. Eventually this was resolved, but for four days I was unable to pay rent or buy food. I was extremely lucky I had friends/family I could rely on in the meantime and that it was resolved so quickly, others may not be so lucky.

I am astonished how many people in this discussion are completely unaware of the idea that some people aren't as lucky as us and work paycheque-to-paycheque. I know HN/SV is a bubble, but surely we're not so out of touch with reality...

I had a very similar experience with BOS back in the early '90s.

And you're spot on - there are so many people juggling these financial balls that one slip up can cause issues for years.

I'd rather keep my money in one place - if you've got 3 accounts in separate banks that means 3 accounts to pay attention to to make sure nothing is going out that shouldn't and also multiplies the risk factor?

So the risk of something happening is larger (but not quite 3x due to shared vulnerability and multiple event), however is risk of being locked and unable to function is significantly smaller as long as you can operate with the subset of accounts.

If you are in the position (like most) where all your account contents are guaranteed and the only thing you are hedging on is convenience vs risk of cashflow problems.

Yup. As long as youre under the limit underwritten by the government it's not a problem.

Unless that insurance policy pays off within hours of your money being gone, it absolutely is a problem.

Sounds like you're describing a mistake there...

Wrote my own JSON parser (https://github.com/MJPA/SimpleJSON) a while ago... not sure how it's a minefield unless I'm missing something?

Perhaps you'd be interested to know that your JSONDemo program fails the following tests:

hang:

y_number_huge_exp.json

segfault:

n_structure_100000_opening_arrays.json

n_structure_open_array_object.json

fail:

n_number_then_00.json

n_string_unescaped_tab.json

n_structure_capitalized_True.json

I would :)

Was going to run the tests myself later but needed a box with python3 on it!

Did you read the post?

The JSON standard(s) are very simple. But in practice this simplicity leads to a lot of edge cases. Very deeply nested structures, numbers that run on to infinity. The point of this post is testing various parsers against these malicious structures.

See this image: http://seriot.ch/json/pruned_results.png

"In conclusion, JSON is not a data format you can rely on blindly." - that suggests the format is bad when it's highlighting problems with the parsers. I see it like saying "Plain text is a bad format because notepad bails on large files"

    In conclusion, NOUN1 is not a NOUN2 you can rely on blindly

This is true for everything. I don't see why this is evidence of anything. If you rely on any system blindly you are doing something wrong.

Is it really an internet wide outage?

Only 2 of the points in the US are affected on https://www.whatsmydns.net/ for the domains we've got on Dyn - same for Twitter etc

If it's under a denial-of-service it's possible that it may respond correctly part of the time.

Since many (all?) of Dyn's authoritative server IPs are anycast, attack traffic is probably not well distributed either. If you're routed to a server that's getting a lot of attack traffic, you're likely to have problems, but a server without much attack traffic will work fine.

"Widespread" might be a better term.

Is this another case of "here's the code I ran" when in fact they didn't? There should be 3 lines of output, not 6!

Also, the code says it will print the time taken since the start of the program, which again doesn't go with the output and the conclusion being made!

Anyway, how come the output isn't in order?

Oops, thanks for that - seems the results from the first run of the example somehow got lost in the final version and I didn't notice.

The order of the output is dependent on when each call finished - they run in parallel, so it's not guaranteed that functions will end in the order they were invoked.

Ah yes, so they do. My lack of sleep is showing!

For some reason I was thinking the readdir would run in series so output would go up by ~1s each time.

Key line before where he shows output...

> However, watch what happens if we double the number of iterations

www.centos.org works for me, but mirrorlist.centos.org (what I want!) doesn't work for me.