That's tough. If it is months, or even (a small number of) years, then there's hope. If it is on the bad side of "if ever" then the question needs to be asked: how likely is it that technology will be capable of giving her a quality of life that she is willing to accept? I have no idea what the legal ramifications are, but since she is obviously conscious and aware, in a sane world her wishes -- based on the best information available -- should take priority over all else.

If the undo/redo history is reliably saved, then yes I agree. But if the application is crashing often enough that you care about this, do you trust the undo/redo history to get saved properly along with the document?

Yeah, that was my initial thought too. But I've already come to terms with the fact that I'm ancient, so it didn't really bother me! ;-)

Agreed. Definitely a case of "please do what I asked for, not what you think I wanted". A properly implemented auto-save feature does not overwrite the original document; it saves a secondary copy, to be used only if the system crashes and you need to recover your edits.

...control-S (XOFF) was used to pause the scrolling on a "dumb" CRT terminal. I don't think I have ever used it to save a document.

Systems I care about (i.e. anything I use for "real work") are on UPSes. If the hardware or software is unstable enough that it crashes unexpectedly more often than once every couple of months (give or take), I fix/replace the hardware or start looking for alternative software to accomplish the same task.

First of all, my condolences. That is a terrible, terrible thing to have happen. I feel especially bad for your sister-in-law, as this is pretty much a worst-case scenario -- conscious and aware, but unable to do anything. The mere thought of being in that kind of state terrifies me.

The brain is quite resilient. Your idea of some sort of brainwave device may actually have some merit; the "biofeedback" craze of the 1970s and '80s demonstrated that you can train yourself to modify your own brainwaves (and other "involuntary" bodily functions), and people have been working on brainwave-based control devices ever since. I'm not sure what's currently out there, but perhaps a creative combination of off-the-shelf sensors and some hacked-together interfaces to a laptop or Raspberry Pi type device could yield some useful results.

If you don't mind telling, what is her prognosis for recovery? Is this believed to be a temporary, or (shudder) long-term/permanent condition? This will certainly affect how you will want to proceed.

If you're primarily focused on meeting the letter of "service level agreements", IMO you've already entered what I'll call "metrics hell" -- a desolate realm where meeting some (more likely than not) ill-conceived measure of "performance" takes precedence over actually helping your users get their jobs done more efficiently. Closing helpdesk tickets within some predefined timeframe is meaningless in the grand scheme of things if you haven't actually solved the users' problems.

Noise and size aren't an issue, as it is in the crawlspace. Out of sight, out of mind... quite literally! All maintenance is done over the network.

All kinds of crap has gotten piled in front of it since I set it up; I haven't needed physical access to the box in over 3 years. When I eventually do need to get at it (which, at this rate, probably won't be until Ubuntu Server 10.04 goes EOL in about a year) I will need to un-bury it first!

I hear you on the Realtek NICs too. While their poor reputation was indeed deserved in the past, they seem to have gotten a lot better over the past few years. I don't know if they have fixed bugs in their silicon, or if recent drivers have implemented workarounds; but whatever the reason, they seem to work reasonably well now.

It is crap like this, and the abysmally unreliable hardware most consumer routers seem to be based on, that has convinced me not to buy consumer routers any more. Been using an old PC (running a copy of Ubuntu Server booted from a CF card) as my router for several years now.

Yeah, I know the power consumption of an old PC sucks compared to a consumer router. But after going through 3 routers in something like 5 years I was sick of dealing with that crap. The PC-based router is way more stable and reliable.

The phrase "technology utility player" almost perfectly describes my current role. My official job title is "Principal Software Engineer". However, I spend less than half of my time doing actual software development. I am also responsible for maintaining the Linux side of our infrastructure (it is a mixed Windows/Linux environment and IT only manages the Windows side); occasionally get pressed into service debugging low-level hardware issues in prototypes of custom boards and cables; and until fairly recently was being shipped all over the country on short notice to support product demos.

TBH I actually enjoy this kind of role, except for one aspect: I don't think management understands that everything takes longer because I am juggling so many different things. All they really see is that Project X, which was supposed to take 80 hours, takes a month or more of calendar time. Or that infrastructure issue Y goes unresolved for weeks because it isn't an emergency (yet) and never floats to the top of the priority list.

There used to be two of us "utility player" software guys here. The other one quit a few months ago, and we were not able to find a replacement with the same mix of skills. This has definitely lengthened the list of things I need to juggle.

Looks like someone can't tell the difference between "March" and "April". Hint: "April" is the one that starts with an "A".

The major distros posted patches for this flaw to their repositories within a day or two of it being made public. (Not to say that it isn't embarrassing for stuff like this to make it into the codebase; but at least it was fixed quickly once it was discovered.)

When you let someone else manage your data, you no longer control your data. That's the biggest flaw with the current "race to the cloud".

Huh. Confirmed. I had always assumed that the normal update process would bring you up to the latest point release within an LTS series. So I learned something today. OP could still install an older point release like 12.04.2 though. All 12.04.x releases are supported until 2017 AFAIK.

You sure about that? The package currently installed on my system appears to match the latest one in the Ubuntu repository for precise.

Well, something doesn't add up then. The OP states the bug was introduced in 1.13.1, yet my copy of Ubuntu 12.04 (which is current on patches) is running 1.11.3.