OK, I fell for it
Your honour, while I am truly sorry, for the sake of the truth I have to add, that I have seen worse rants being ment word-by-word
136 publicly visible posts • joined 9 Jun 2010
I recommend you do exactly that. Never ever again use a bit of open source software for you personally or for the good state of Texas.
That will show us "commies" and "libtards" what we need to know and make your IT run smoothly on "highly secure quality products from the likes of Microsoft, Apple, Oracle". I do especially recommend the latter.
In the Joules/Gigainstructions performance indicator, which is what matters most where MALI-GPUs are used, they have a 1:10 advantage over Intel and still more than 1:5 on (today's) Nvidia.
While they are good enough for what they typically are used for, this is one of the areas where the merger could actually generate better products. Not a good enough argument for me, though.
One can think so or so of Ubuntu, but the Nvidia stuff is just an "apt install" away.
We do lots of work for the broadcast industry, that includes lots of NVDEC and NVENC accelerated de- and encoding and lots of CUDA-based image filtering.
The combination of Ubuntu Server, the in-repository nvidia drivers (and even the out-of-the-box NVxxx-enabled ffmpeg) make this really pain-free. Definitly less painfull than a Windows 10 (or worse: Weindows 11) installtion.
I petition the Register for a new unit of measurement - for moronity (this is the evil flavour, let's use "idiocy" for the foolish-but-not-evil flavour)
I propse the natural choice: The Donald
On such a scale, this would come up to nearly 0.1D (or just shy of 100mD)
A lot of us use Linux - including the overwhelming majority of servers. And the Linux drivers license allows you to drive both, a Ford and a Ferrari with the same ease.
Now if some clever big chap were to understand, that it is not Intel lagging behind Apple, but Intel lagging behind Arm and do what Apple did for Laptops, but target the Linux server market, this would be quite a Bugatti.
SQL server in itself is one of Microsoft's better products. But.
- The possibility to run it on a sane OS came years after the others ate this part of the cake
- Licensing (for use on your own infrastructure) seems to be modeled around Oracle: Nobody understands it.
So thanks, but no thanks.
What I have a hard time understanding is the unreflected "must charge in 15 minutes" argument. It may be true: What percentage of cars is actually used in a role, where a range of say 500 Km / day isn't sufficient with an 8 hour overnight top-up?
I am in the process of selecting an EV to buy, which implies, that I did a thorough analysis of use-cases. My personal cut-off is below 300 Km / day, and I know not of a single person, where this would be beyond 600 Km / day.
Combine parked-time or overnight charging with some smart grid-balancing by means of a charging priority (e.g. I don't care if my car is full at 2 a.m. or 6 a.m., but I want it to be guaranteed full at 7) and you might get quite a tasty package: Grid-reserve is maintained by switching off thousands of chargers for those 30 minutes it takes to get that gas turbine online when something big fails, instead of idling it all day or all night.
Call ne sentimental, but I was quite attached to IBM for many years. Their x-Series servers were solid like rocks, and at the time this was even more important than it is today. And they were quite early on the "Linux is the default OS for a server" bandwagon. The SCO saga (most of you will remember groklaw) added more respect to that.
But it is not ment to be. They seem to be just are unable to cope with the realities of today, where you can't slap an IBM sticker on something to double its perceived value, where the only ultra-high-margin cashcow remaining is the mainframe, which is going the way of the Dodo. Any of those perky new banks or insurance companies using them? Nope. And DB2 neither.
Instead of adapting, they chose to do more of the same - not good. Bye, IBM, it was nice while it lasted.
Oracle seems to have managed to let the licensing problem stand long enough to make it irrelevant: Scaling file systems has gone through a few paradigm changes in these 20 years, so that the target market for ZFS is ... not that big any longer.
Add to that some arcane rituals needed to be performed if you run into one of those nice little 'cannot mount ZPOOL' siutations and it is just no longer relevant except for some rare edge cases.
I'd have happily invested in it 20 years ago. Even 15. Buto today? Nah.
We all know the "This website wants to send push notifications, Allow yes/no" prompts. It has proven quite a well-working way to deal with site privileges. Nothing (ahem ... nothing technical) would stop browser vendors from defaulting all such APIs to disallowed, but present the user with a "This website asks for these privileges: 1., 2., 3., ..." dialog when first loaded,just the way it is with Apps from some App-store.
PWAs that actually put value into the users hand would have a good case, but not every click-bait site.
Removing the "secret" from "secret sauce" seems much saner to me than disallowing sauces alltogether. It would also help to move things from the "App"-Domain back into the "Open Web"-Domain, which is of cause what Apple fears like a Vampire fears daylight.
We do support a large number of appliances in really remote (to us, as behind doors with gun-carrying guards around them, who always - and I really mean always - look annoyed) places.
Of course they are Linux based.
Of course they shit themselves from time to time (TBH basically only with hardware faults)
Our "Fleet Management" is a device/software/version-personalized script on the appliance that phones home every so-and-so telling e.g. "I am fine" or "I'd rather have a new disk in slot 3" (they carry a largish cache each, but no genuinly original data apart from the system-generated config). And that "phone home box" is of course a cluster and of course keeps track of missed calls.
The "dashboard" is ugly as hell, but nobody ever bothered to have the designers put a lick of paint on something, where the "show last 100 events with lever warning or worse" logtable is the page that gets 99% of the hits.
If such an event occurs from a not-yet signalled device-id, a list of e-mail addresses will be pinged once an hour until they log in and display details for this device id.
If they log a warning+ event or if ordered to do so when phoning home, they will also open a reverse SSH tunnel to the watchdog server, so that the admins can log in.
The whole setup took less than 2 days to create and hasn't failed us in many thousands of deviceyears.
And: It is no subscription, belongs to us, isn't tied into some "cloud service" and moving it is a DNS record away.
And: it is ISO123456789-certified to exactly 125ml of Pinot Grigio
... of "physical access = complete compromise modulo time".
And this tends to be even cheaper and easier with "Smartphones" (i.e. smallish pocket computers) than with the PC counterparts - and I know quite a lot of people, who use the former to unlock the latter.
Some sanity might just help: This document is important? Encrypt it on a file level and PLEASE close it before letting you computer unsupervised.
On the opposit end of the spectrum, they might have all my "photos" folder. Well, the Memsahib might take issue with that, so not really all.
... a cronut at your desk without worrying about subsequent trips to the Genius Bar.
I always could. Since I need my notebook for work, not to show of wealth and/or moronity I didn't and wouldn't dream of buying such a device (and at such a price) for the privilege of running OSX "Vista" Catalina.
... for anyone thinking he actually owns that expensive gadget he shelled out for.
And a briiliant burglery reconnaissence tool: If after the 17th loop of "Last Christmas" (or any title in the 1000 volume collestion "Songs I'd have hoped to never hear again") nobody is running out of the house with blood dropping from the ears, the target can safely be assumed as empty.
Full disclosure: My first language is German, where the word "Schadenfreude" comes from.
It consists of "Schaden" (loosly translatabel to "damage") and "Freude" (loosly translatable to "joy"). So "Schadenfreude" is enjoying somebody else's damage.
Now I strongly reject the notion of me really enjoying something, that only damages Facebook! I will not settle for anything else than full destruction. My (bohemian in both senses) great-grandmother would have said (again loosly translated into modern english): They have already sinned more, then what can be counterbalanced by a single eternity in hell.
Let's start with the factual opinion: The important part in today's AI workloads is not so much a question of performance alone (you can throw a lot of hardware against it), but of performance per unit of energy consumption. It is rather hard to believe, that even the smartest trick can make a general purpose CPU work as efficient on such a specialized problem than a specialized (i.e. GPU) accerlerator card.
Now the meta opinion: Whatever is bad for Facebook is good for me. YMMV.
Of course we see from the enlightened example of our Trumpian friends, that draconian punishment (including death) are a near guarantee for a very low-crime environment, peacfull streets and no cybercrime at all.
Speaking of it: Are you interested in an investment opportunity into an essential and lucrative bridge?
Also from the FA: "However, LVI turns out to be most practically exploitable on Intel processors because of the combination of the facts that we have seen more Meltdown-type leakage sources there that can potentially be inverted, plus certain design decisions that are specific to the Intel SGX architecture (i.e. untrusted page tables)."
Quote-picking doesn't alter the real world.
First of all: I appreciate StackOverflow
This stems to a big part from those nasty bastards that pay my bills: The customers. A lot of our work is replacing outdated/buggy/no-longer-suppported/whatever systems and applications with something more robust that runs on a sane foundation (for varying values of sane). This often includes having to dip into the other system deep enough to make migration easier or sometimes even feasable. StackOverflow has this knowledege.
As a thank you, I tend to answer a selected "hard" question from time to time - quid pro quo.
Now the second flame war: There is no inherently bad programming language - and if there is, then neither JS nor PHP. It might be (and likely is) true, that the perceived easyness and accessability of the pair produces an out of proportion amount of inexperienced or just bad developers, but I consider it unwise to blame this on the languages. We create a lot of PHP - but we (at least try to) do it in a sane, controlled and responsibe way.
I do on the other hand know from personal experience a metric shitton of C{,++,#} code that should never have been allowed to leave the dark cave it was conceived in. And don't get me started on 100MB+ Java processes doing basically a "Hello, World".
Languages (and more important: Runtimes) are tools. Tools help. Tools do not replace skills.
This is neither cynical nor a flame war, but an honest question:
Why would anybody use VMware for virtualisation in more than a Lab/SMB scale?
- If you are so big, that you feel the pain on 33+ core CPUs, you most likely have an IT department
- If you have an IT department, the actual people working there will want the money going in their pockets, not VMware's
- Xen and (even better) KVM cost nothing in license fees and scale much better, the HCL of KVM being ... close to everything ever built on x64.
- Just a handfull of VMware full-featured licenses will easily pay for a BOFH plus a PFY.
- The "ecosystem" (managability, storage options, ...) of the FOSS world have already bridged the gap and are often better than their VMware counterparts.
We take a fundamentally different approach to tooling - I understand, that this causes more maintenance effort but the resulting increase in productivity and happyiness ahs until now outweighed it by far. It may be hard to scale though.
The "normal" stuff:
For every project
- there exists a defined repository (duh). What's not in there dosen't exist.
- there exists an automatic build and deploy process to a defined set of targets (usually VMs, that are spun up from a template). What doesn't build there doesn't count.
- Deep testing in development is encouraged, but only the tests on the targets count
The "strange stuff":
Every developer choses whatever tooling he wants to fill the repository. Choise of Hardware (within a budget), OS, IDE, tools whatever are completly up to the developer. Want to develop in nano via SSH? Go ahead. Want Visual Studio? Go ahead. Use whatever makes you happy, but remember that the build target isn't your machine, but the project build-and-deploy bot.
Some chose to "make things run locally" first and then basically port to the target, others keep very close to the mothership - I don't care. It's the output and dev happyness that count.
A few days ago I was asked by a client (for whom we normally don't do any networking, but hey there is the water-cooler/coffee-maker/whatever talk) to help them think how to expand their (currently Ubiquiti) WLAN - stay with Ubi, roll their own or go "Enterprise". I forwarded them a link to this article at ca. 11 p.m., and got a reply along the lines of "so one possibility just dropped" within 5 minutes.
This tells me, that they are quite serious in both ways.
Yours humbly is from a country that is better known for its musical than for its technical prowess and whose biggest telco was at that time a mostly state-owned partner of Vodafone (now a daughter of America Movil after a more than boitched privatization job).
Their claims of 95+ percent availability seemed rather absurd then, about 90 was what everyone would have guessed (remember that this means double the downtime).
Now: The clue for the why is in the first paragraph: It was a mostly state-owned telco. Management was picked by their party affiliation and willingness to do what some politician wanted, not by what was best from a technical perspective.
Let's face it: Using a public cloud offer always comes with a measure of lock-in: A spectrum from just "convenience lock-in" where the effort of moving a workload to another provider or self-hosting is just a resource-costing nuissance up to a "systematic lock-in" where you need to rebuild a lot of components to make it even possible to move to a different vendor. We have come to accept this and many (if not most) mitigate as good as possible.
The only reason we are willing to even consider accepting it is an uneasy trust not in the vendors, but in the market forces moving them: If one cloud vendor were to start screwing his customers more than the pain threshold, he would have a very hard time acquiring new customers and stall his growth - which is very near to a death sentence in an industry where economy of scale is a major effect.
There is one thing though: For many of us (and this definitly includes me) this trust does certainly not extend to Oracle, whose business model is sometimes perceived as "lock, screw and blackmail". And while this does work with (mosty legacy) applications that are tied into the main Oracle Database product, the idea of beginning or extending such a relation to cloud hosting does not go down well with IT departments. And forcing a product into goverment contracts via the courts doesn't inspire confidence as well.
So Larry made a cloud, and nobody wants it - if not coerced with one of those beloved audits, basically nobody signs up. Beware of a furious and desperate Billionaire!
There have been many (both enlightened and whining) comments in the last years that Microsoft has basically given up on the SMB (and maybe "Power User") market:
- No more Exchange in SBS
- No opt-out of "Telemetry" below hundreds of seats
- Mo more WIndows for mobile
- More of that stuff.
Well, IMHO this is nothing else than completion of the circle: If we don't care for the SMB market, then why would we care for those supplying it?
What I am quite curious about is, whether there is some vacuum starting to establish and how it is going to be filled.
First of all: I don't game, so the blind is speculating on colours here.
But: I assume a decent gaming rig will cost over 1000,- (Euros, Dollars, Pounds doesn't really matter) - let's say 1200 to ease up the calculations. This means, that with a subscription cost of 50,- the hardware alone will cost two years worth of it, not taking into account the power it uses at your end (a.o.t. on Google's end). Most hardcore gamers I know will consider a 2 year old rig hopelessly outdated at least concerning the GPU. Add to this the cost (in expertise and time) of running Windows 10 and outsorcing all that to bigger players (pun intended) and only owning what is basically a smart-TV-sans-TV might actually be a good value proposition for the less tech-savvy.
Far be it from me to defend intel. Far as in at least a few galxies.
But in the name of fairness one has to make clear, that Nvidias GPUs have never undergone such scrutiny - and they would have fared rather poorly if they had: Just the NVENC part (which makes up a tiny proportion of the GPU) leaks the last image of every encoded stream to any Dick, Tom and Harry who create a new context.
Canada is part of America (some would argue the most sane part of it):
Canada is NOT a part of the USA (which of course explains why it is often seen as the sanest part of America).
As of 2019 the correct wording is "Look, boy! Now I'm in Trumpistan! Now I'm in Canada! Trumpistan! Canada! Trumpistan! Canada!".
On a more sobre note: Please don't help spreading the "fake olds" that "America" and "USA" are synonymous.
Since yours truly lives in Vienna let me elaborate: Vienna City services (from trash collection to dog tax) are organized into "Magistratsabteilungen", or "Divisions of the Magistrate", called MAx with x being a Number between 1 and somewhere around 70.
The series is about an imaginary "MA2412", the division responsible for Christmas decoration - you can imagine that a division of civil servants having exactly nothing to do for 10+ months of the year (and I mean exactly nothing, not the figurative-but-true nothing) creates a good backdrop for a comedy series.
Bad thing though, that neither script nor acting nor ... ahem ... anything in that series is of better quality than what would be produced by a "Magistratsabteilung"