I'm one of those people who prefers to converse in person and will tend to ignore requests to "follow up" by email.

I've found sometimes it's the "follow up" people who are backstabbers. They will do such things are talk with you at length about an issue, and then post to everyone the same question in such a way as to imply you never had the conversation, making everyone else think you're not willing to share information, just so they can force you to do it all over again, on their terms.

Is it a few minutes to summarise? No of course not, it's half a day of writing to meet their standards.

Is Documentation the key to eliminating confusion? Well good documentation is!

Misleading documentation, as every programmer who has wasted their time believing documentation and comments instead of code knows, is worse than no documentation.

But we were talking about email. It's the lack of structure while using it as a "but you said X in that thread 1,123 emails ago" stick that I find difficult.

Where I work, a lot of things are doing by email. So much of importance is written in emails,m yet *nobody can find things or keep track of things because there's so much email it's impossible for anyone to remember*. Even the most pro-email, pro-documentation and pro-structure people (which I respect) tell me they don't keep up with it, while simultaneously being advocates for it. That's not good documentation and in that scenario, those email trails seem to be more of a tool for ensuring that little productive work is really possible.

That stuck out for me. Depending on what you include in those 50kloc, that may be a relatively small project, compared with the ones being discussed in this article.

In my experience, those projects where new things are constantly going wrong in production are those with a large number of slightly flaky and somewhat complex dependencies, 3rd party packages, black box libraries, other products/components they have to work with, that themselves need to be configured/built/patched etc and themselves suffer from production problems, and nothing ever works quite as anyone expects. Counting the full size of the dependencies causing fragility would take it to Mloc scale.

At 50kloc scale, especially if that's all your own code, that's within reach for one person to understand enough of the project at a time to keep all the parts in good working order.

Or you might be talking about 50kloc projects with Mloc-scale dependencies that works out pretty well, in which case I salute you :)

I have a couple of smaller, cheaper laptops with Windows 10 on that keep nagging to apply a pending update.
They are only a few months old.
But telling them to proceed just results in them downloading about 5 gigs data, and then giving up because they don't have room on their small SSDs.
This has happened several times, and is a pretty big waste of bandwidth.
We have put hardly any non-OS data on them. They just can't handle the updates, out of the box.

Does this chance mean they might be able to finally update?

Where I live you just hold your card in front of the reader for half a second.

No sliding, no PIN, not much wait. Definitely faster than counting out change in cash.

Wrong. This type of attack targets known IPs of public servers, directly or indirectly.
A bigger subnet doesn't help.

It's because Docker is written in Go, and it's one of the most high profile Go projects.

http://www.slideshare.net/jpet...

Perhaps it should, but you're describing a router that doesn't comply with DHCP standards and asking iOS to workaround the broken router.

A compliant DHCP server will remember DHCP leases across a power cycle.

Good points.

Unfortunately, if I understand correctly, most of the publicised password breaches aren't due to intercepting the password in flight; they are due to successfully finding plaintext that matches the salted hash in the stolen database.

But still that's an interesting idea for protection against intercepting passwords in flight or recovered from RAM later.

Not only are passwords in plaintext in-flight behind a TLS terminator, typically on the LAN or virtual LAN or unix socket going to authentication services, but they also linger in kernel, router and middleware memory long after they have been transported around. Good security software knows to erase its own memory of sensitive data after use, but in between the TLS-terminating load balancer or web server (NginX, Apache etc) and back-end application processing of security URLs, usually there are a great many subsystems that transport the data of HTTP requests in plaintext, and don't have special provisions to erase their memory of them. So they linger in RAM afterwards until randomly overwritten. This can be made more secure than it usually is, but it's quite advanced.

The encryption part of your idea can be implemented today by sending Javascript to the client to encrypt the password.

But unfortunately if there's a real-time breach on the server, the attacker can probably change the Javascript as well.

Noise cancelling headphones/earphones need power and have their own batteries. Some for those with mics in for making calls without removing them, or where they use a DSP to, hopefully, improve the sound of imperfect drivers. With lightning they can use the phone's battery.

That said, they could have done the same over a backward-compatible 3.5mm jack if they'd wanted to.

When they started talking about "courage" in the keynote, I got a little excited that maybe, just maybe, they meant they were couragously opening up Lightning for their competitors to use and make it a universal standard.. But no. That would have been courage, imho.

Last time I checked, yes it did send messages to Google, even running Chromium, on Ubuntu Linux, with all the undocumented command-line options I was able to find to disable various functions. That surprised me.

I was wanting to use Chromium to view local applications without any inappropriate network traffic; it wasn't suitable.

LibreOffice is great, and some of its tools (I'm thinking of the change history) are in some ways better imho than MS Office.

But I wish it was more reliable. I had to subscribe to MS Office recently because LibreOffice (even the latest versions) was corrupting images in documents and screwing format around them (unrecoverable once saved), and mis-displaying basic highlighting in even very simple documents wth nothing obviously fancy. As these were contracts and things like that, some of my colleages were getting the wrong information, with quite serious consequences.

The first problem - corrupting images - has no excuse. You shouldn't be able to edit inside LibreOffice, save, load, and get back something different, no matter what quirks of file format compatibility there are.

I had to not only switch to MS Office to read documents sent, and edit documents to send, but I had to work out which branches had been edited by a colleage with LibreOffice and find the parent version that wasn't corrupted to redo all the edits on that branch.

I say "colleagues" but I'm talking about a non-profit, where I don't get paid and neither do they. To save money, my job now includes "editor of important documents like that" just because I've got "the" MS Office licence :/

I've been using mine daily since they were first released, and it doesn't crash or hang.

It's slow to start apps, and due to lack of updates the browser is no longer compatible with some current sites, annoyingly including some that used to work fine.

I'm in the market for something more modern to replace it, but like the GP, still haven't found an appealing replacement.

Android makes me uncomfortable due to the Google-mothership thing, or any other proprietary service for that matter,
I know you don't have to, but at every turn it practically begs you to link up with them to use it properly, and it doesn't
come with options to link to your own generic services instead.

I have an Apple laptop but the iOS walled garden is even less appealing.

But hardware keyboard is probably the #1 feature I'm looking for.
An Android 4g device with one of the current top quality cameras, a fast CPU, enough storage, and a hardware keyboard (that doesn't replace half the screen - sorry Blackberry) is probably what I'll end up with - when someone makes one.

It reminds me of the harmonic drive - a low backlash, high ratio compact gear.

Other comments have noted that a very high ratio would need very strong matariels to transfer significant power.
That's true, but sometimes the point isn't power, the point is to move things over very small distances precisely.