Posts by thosrtanner 287 publicly visible posts • joined 7 Mar 2013
The author is going to be very happy right up to the moment all his agents decides one of his machines need to be used as a sacrificial data source.
We all know what AIs can do when they get it right. And now we're discovering just what they can do if they gut it wrong. Which - clearly is not a concern for some.
is this just windows for organisations or just windows for ordinary people? I've never used windows backup because there are better solutions out there. I lose the registry settings, which is a pain, but restoring your registry settings without restoring all the programs that set them up is a bit flawed. And the registry is horribly flawed. Almost any other method of storing settings is preferable.
What happened to the Atom/RSS feed for the BOFH pages? That appears to have disappeared with the new site.
Oddly there's still a feed for forum questions that's visible in the browser (if you have a browser that shows it in the address bar, which mine does). If I look at the page info it shows
* Posts in this topic
* The register: whole site
* The register: Site news section
as feeds, though the target of 'the register: whole site', which I was using prior to this points to a completely empty page. It should really have been redirected to the new feed page.
Additionally the BOFH page doesn't have any appropriate header/metadata info either and the URL i used to use, https://forums.theregister.com/offbeat/bofh/headlines.atom, seems to have a broken link for the last bofh entry.
Please could you fix these
I think UAC appeared in windows 7 (and TBH I didn't find it too annoying. Not nearly as annoying as the software that updates itself without warning and puts itself in your appdata directory where anything can mess with it). Windows 8 was reviled because giving a desktop PC a UI that made it look like and behave like a tablet was absolutely insane.
palemoon browser still tells you if a site has an rss feed (if it is included in page data) by putting an rss icon in the address bar. This works fine for the Register (and quite a lot of other sites, actually), but not so fine for the BBC who are definitely members of the locked cupboard in the basement with a beware of the tiger sign school of communication. I've been using palemoon and the inforss browser extension (which I maintain, so I may be a little biased...) to get myself a nice news ticker at the bottom of the browser window for yonks.
It's all very well complaining that "I only changed one line in a function that nobody uses so dependandabot shouldn't alert everyone", but - how is it to know that you only changed one line in a function that nobody uses without doing a thorough investigation of both the changes to the library, presumably building a complete call tree, and the (implicit or explicit) uses of the library in the client code as well.
If the complainant is that fussed, maybe he should produce a version of dependabot that does all that, and offer it as a replacement. I'm sure it'll run super fast.
I'm quite happy with "this thing you use has changed. rebuilding your thing is probably a good idea" alerts.
And in answer to one of the other "only update dependencies if your code has changed". Um. Why? What about when the library you're using has fixed a security issue in code that you use (possibly without realising, because frankly, most people have little idea of what goes on underneath the hood in libraries they use).
If you want to minimise your alerts from dependabot, you should not use external libraries. Obviously it's going to be a lot more work for you, but you can rest assured that the number of alerts you get will be much lower.
I remember at Cambridge (UK) undergraduates had allocated online 'tokens'. The charging rate went up at busy times of day. We very soon learnt that attempting to compile an algol68 program at lunchtime burnt out your online tokens in about 15 minutes. As a result we did most of our online work after tea before disappearing to the college bars (which at that time closed when they felt like it rather than the closing time enforced in pubs. and of course the college bars were somewhat cheaper).
People have been the problem since for ever. The Roman emperors found out that bread and circuses kept the masses happy and they (or at least the ruling classes) could do more or less what they wanted. That hasn't changed much in the last 2000 years. The bread is now fast food and cheap booze and the circuses are provided by the media/social media.
It's the "You must open an account in order to pay for something" bit that really annoys me. I'm AT the place I'm purchasing from, you do NOT need my email or my home address in order to make a delivery because I'm right here. Why should I need to open an account?
I can't help feeling that a class called SoapHttpClientProtocol accepting anything a target that explicitly says not to use Http or Https is a bad idea. Writing to a file is not using http.
So yeah, I sort of agree with the researchers on this one - it should either error or ignore the protocol part - preferably the 1st.
unfortunately "localhost" is just a hostname that by convention maps to 127.0.0.1 - if you happen to edit /etc/hosts and put a different value for localhost in there, you'll get strange behaviour on your system. Especially if there is no server at the target.
It's be interesting to know if they broke 127.0.0.1 or just (I use the word loosely) localhost.
banks make money from the paltry sums in your current account as well. not to mention the time it takes for money to transfer between account. it mysteriously takes 3 days for cheques to clear, during which time the banks are making money - according to a lecture I went to from the BACS people the underlying systems can do the clearing overnight
Even Stroustroup makes mistakes. His reason for not having exceptions checked at compile time was that
float mysqrt(int a) { return sqrt(a < 0 ? -a : a); }
would never throw an exception but the compiler would insist it could and might refuse to compile code that used it and didn't state they might throw the appropriate exception.
mysqrt(MININT) anyone?
Also, to be frank, that's such an exceptional case I'd rather put the try/catch in myself.
Pretty sure that cont pointers and const refs are pretty close to viral annotations - const correctness is a PITA if some low level thing guarantees not to mutate something it is passed in the documentation but not in the API. Especially those things that have APIs where the mutating and non mutating version of the code are differentiated only by a passed flag.
Although the authors are careful to say "it's not necessarily discrimination", it does seem to be a particularly pernicious form of discrimination - using a particular AI says nothing whatsoever about the candidate and their ability to learn different techniques. we appear to be moving from ageism to AI-ism
So in my first job, I was put onto doing the programming for a completely programmable video display - so programmable you had to have a program loaded into its rom for it to do anything. And one of the things they did was decided to have a removable keyboard.
Obviously a terminal without a keyboard is going to be hard to enter data on. So, I made the code able to detect various conditions including keyboard being removed/added. On removal it flashed up on the top line "give my keyboard back, you thieving basket" and then replaced that message with "thank you" when a keyboard was detected, which message went away after a short while.
This was apparently so impressive it got demonstrated to customers as a feature of our system. and it was several years before someone complained about the content of the message!
