It appears we are back...for how long is anyone's guess.
BD139 wrote a couple posts back something to the effect of "all forums suck". I agree. They are at best "tantalizing UI enhancements" to communication problems that Usenet manages better.
Usenet was even worse due to the whole eventual consistency thing.
I've sort of resigned to the fact that eventual consistency is the only consistency is we're going to have, at scale. Better then to grasp that critter by the horns or scrotum and deal with it. Which is going to suck, of course.
Yes, I believe that too. Speed of light isn't infinite, there's no such as global time, and the limits of sequencing were defined decades ago by Leslie Lamport.
If you believe otherwise, I'd like to see what you believe will be the effect on the limitations of two-phase commit protocols - and when three (etc) phase commit protocols are necessary.
Basically parallel processing sucks in ways serial processing doesn't.
Naturally the hardware mob have had to address this inside processors, e.g. superscalar processors, speculative execution, cache coherency etc.
There's another way if you change how you look at the problem from a partitioning perspective. Smaller the partition, the quicker consistency can be resolved as a rule to the point you can have transactional consistency. The trick is to redefine the partition size based on the operation in question and the participants rather than the current lazy arse cheap thinking which is have one partition with all participants in it. This is complicated and specific to every single problem domain and not generic which is why you don't see it.
There are fundamental that are common to all domains, since the problem cannot be eliminated. It is constrained by communication latency (including lost/repeated messages) and processing time. Partial workarounds are, as you point out, possible but are very application specific.
The hardware people have that problem inside a single IC. RF people have an equivalent in conductors
Got these Lattice analog FPGA !!!, ispPAC10, what should I do with them if I can program them ?!?!
Cheers,
DC1MC
Man those are ancient. Lattice, to their endless credit, still have downloadable dev software for those things, long after the devices themselves were shitcanned.
https://www.latticesemi.com/Support/MatureAndDiscontinuedDevices/ispPAC You'll need to have a Lattice login, or register for one, to actually download this.
[/quote]
Thank a lot for the link , like many old or well made windozian software it runs perfectly under Wine (except of course the dirver for the programer, but there the Xilinx JTAG player is a big help).
Also a big
Danke for the
@ch_scr for the assist.
The ispPAC80/81 were really capable devices, too bad they've canceled them
-
If anyone in DE needs a couple to play with (looking at you
@capt_bulshot as well
, please let me know, you can't bet the price
.
Screenshot related.
Cheers,
DC1MC
I'm seeking opinions please. I'm looking at replacing my SSHD drive in my EliteBook with a proper SSD, and I'm thinking about this one Samsung 860 EVO 500 GB SATA 2.5 Inch Internal Solid State Drive (SSD) https://www.amazon.co.uk/Samsung-MZ-76E500B-EU-Solid-State/dp/B078WQT6S6/ref=sr_1_4?dchild=1&keywords=SSD+500GB&qid=1620478878&sr=8-4
Its the only that comes with a 5-year warranty on it, the others are 3 year and some don't state what they are. Anyone ever had to call on a warranty claim with Samsung before? If so, how was that experience?
We’ve got over a hundred of these drives and later ones in laptops and desktops. I’ve used them for about ten years. Never had to do a warranty claim - they are absolutely rock solid.
Bah, Amazon are out of stock till June Argos however aren't and secured my order, just about to install and transfer data etc across
Yeah Argos are quite good these days. I had to order something from them the other day because Amazon were out and arrived 2 hours later! . Enjoy the SSD - they work flawlessly.
Just got back from the walk. Was absolutely crapping it down all day but was a good day out despite being asked to hob it off a hill by some very nice and friendly armed police officers . Chequers estate for ref https://en.wikipedia.org/wiki/Chequers
Inevitable photo up a hill. Didn't realise it was quite as sodden up the top until I got up and realised I had a wet arse
Yeah, I already have a SSD in my desktop, had for years now, and it performs flawlessly. Once I cloned the old drive on the laptop, swapped the drives over, it was chalk and cheese, can't believe how slow Windoze 10 was running on it, it's an Intel i7 running at 2.67GHz, but that SSHD was running more like a conventional HD FFS. I’m pleased to report now it is much faster responding and will see more use as a result.
What on earth possessed you to attempt to go walking anywhere near Chequers, did you have an alternative motive? So you got a wet soggy arse and then decided, after being asked to leave, to go home and break the bleeding forum because you don't have a cat to kick tut tut
I hope the EEVBlog Forum Server's location is not on Maldives, there might be a "correlation" with the downtime
This video might be fake, but looks scary if the Chinese rocket reentry was really filmed
Bottom right says 38S 145E which is Australia, not the Maldives.
Bah, the HP 8570A is proving annoyingly difficult to find info on, probably not helped by HP having made a laptop with the same model number
I find google are becoming more & more useless for finding info on TE, they just bring back loads of modern crap with the same part number, we need a better search provider that isn't constantly trying to push new online shop stuff.
No dead tree editions on evilbay?
David
I'm not even going to attempt to catch up on this thread, but thought I'd drop in to say howdy. I've had a few things follow me home recently. A friend of mine gave me a Type 454 that he had lent to another friend and it blew up on him. That repair is still in progress. The +75 volt regulator failed short and let the +150 unreg onto the +75V rail and killed a bunch of transistors, mostly in the vertical output amp, but also a few in the preamps. Also of note, this failure caused the HV to totally wig out and arc violently inside the HV cage, around the output to the anode. Replacing the failed regulator fixed that fortunately and nothing was damaged in the HV.
I also recently had a NICE Type 547, cart, many plug ins, and manual follow me home from an estate sale. Just did a little touch up adjustments to the time bases and the sweep mag registration, and she's been put to work in the lab!
Welcome back and that 547 looks real good. I'm jealous. I'm still fighting with mine. PSU issues.
Thanks, it's a real sweet scope. I believe the case can be made that the 547 (and it's dual beam big brother 556) are the finest cathode ray laboratory scopes out there. Jim Williams certainly thought so: https://www.edn.com/be-it-ever-so-high-tech-theres-no-place-like-home/
What's still ailing yours? BTW I got this one off the cart and on it's side to look at the underneath...looks like someone before me replaced a few of the Si rectifiers, much like I had to do with the 585A. Very clean work though, and the rest of it is pristine looking.
The +100V supply keeps smoking itself and I'll be damned if I can find the source of the short. The -150V reference is also low.
I may be PM'ing you to take some resistance measurements for me.
There was someone in the UK with a 585 that kept blowing fuses when the time delay relay energised, a number of parts on the +100B rail had failed too. It turned out to problem with the mains transformer, he did eventually manage to get e a replacement from another forum member.
https://www.vintage-radio.net/forum/showthread.php?t=153491David
There's another way if you change how you look at the problem from a partitioning perspective. Smaller the partition, the quicker consistency can be resolved as a rule to the point you can have transactional consistency. The trick is to redefine the partition size based on the operation in question and the participants rather than the current lazy arse cheap thinking which is have one partition with all participants in it. This is complicated and specific to every single problem domain and not generic which is why you don't see it.
But that is what Usenet is -- the hard consistency is only enforced in small domains, which produce atoms (RFC 5536 messages), who once committed are possible to refer to from other atoms, and you can create a thread of messages that constitute a discussion. This thread construct is possible to build in an asynchronous way, and it will never be perfect, but it will approach completeness with increasing age of atoms.
Yes, that is the happy puppy view of distributed systems consistency, but there is something to it.
Bottom right says 38S 145E which is Australia, not the Maldives.
That's true... therefore we have to wait until the authenticity of the video is confirmed by independent sources :-)
At the moment the Internet is full of rumors, viral videos and false information
Once the Rockets are up, who cares where they come down?
That's not my department, says Wernher von Braun
By the way, Wernher von Braun obviously learned Chinese (as sung by Tom Lehrer few decades ago)
Bah, the HP 8570A is proving annoyingly difficult to find info on, probably not helped by HP having made a laptop with the same model number
never had my hands on an 8570.
but my recollection is that it was just an 8569b with no provision for external mixers and no resolution bandwidth below ummmmmm maybe 1Khz?
ko4bb has plenty of stuff on the 8569b.
edit oh yeah.....1khz......there it is.
https://accusrc.com/uploads/datasheets/6252_8569B.pdf
There's another way if you change how you look at the problem from a partitioning perspective. Smaller the partition, the quicker consistency can be resolved as a rule to the point you can have transactional consistency. The trick is to redefine the partition size based on the operation in question and the participants rather than the current lazy arse cheap thinking which is have one partition with all participants in it. This is complicated and specific to every single problem domain and not generic which is why you don't see it.
But that is what Usenet is -- the hard consistency is only enforced in small domains, which produce atoms (RFC 5536 messages), who once committed are possible to refer to from other atoms, and you can create a thread of messages that constitute a discussion. This thread construct is possible to build in an asynchronous way, and it will never be perfect, but it will approach completeness with increasing age of atoms.
Yes, that is the happy puppy view of distributed systems consistency, but there is something to it.
That’s ok until someone doesn’t get an ancestor message. Which is where the consistency guarantee fails.
Im working on something in this space at the moment. Targeted at contract resolution.
I'm not even going to attempt to catch up on this thread, but thought I'd drop in to say howdy. I've had a few things follow me home recently. A friend of mine gave me a Type 454 that he had lent to another friend and it blew up on him. That repair is still in progress. The +75 volt regulator failed short and let the +150 unreg onto the +75V rail and killed a bunch of transistors, mostly in the vertical output amp, but also a few in the preamps. Also of note, this failure caused the HV to totally wig out and arc violently inside the HV cage, around the output to the anode. Replacing the failed regulator fixed that fortunately and nothing was damaged in the HV.
I also recently had a NICE Type 547, cart, many plug ins, and manual follow me home from an estate sale. Just did a little touch up adjustments to the time bases and the sweep mag registration, and she's been put to work in the lab!
Welcome back and that 547 looks real good. I'm jealous. I'm still fighting with mine. PSU issues.
Thanks, it's a real sweet scope. I believe the case can be made that the 547 (and it's dual beam big brother 556) are the finest cathode ray laboratory scopes out there. Jim Williams certainly thought so: https://www.edn.com/be-it-ever-so-high-tech-theres-no-place-like-home/
What's still ailing yours? BTW I got this one off the cart and on it's side to look at the underneath...looks like someone before me replaced a few of the Si rectifiers, much like I had to do with the 585A. Very clean work though, and the rest of it is pristine looking.
The +100V supply keeps smoking itself and I'll be damned if I can find the source of the short. The -150V reference is also low.
I may be PM'ing you to take some resistance measurements for me.
There was someone in the UK with a 585 that kept blowing fuses when the time delay relay energised, a number of parts on the +100B rail had failed too. It turned out to problem with the mains transformer, he did eventually manage to get e a replacement from another forum member.
https://www.vintage-radio.net/forum/showthread.php?t=153491
David
The transformer appears to be OK. Proper AC voltages.
Note the US Navy asset label... there's your explanation for its battered visage!
Also welcome to the enablement thread, hyl
If massively battered stuff is your prefered challange:
Sometimes there are lots of HP devices on the german bay system posted by Israelian sellers looking like beeing fished from military dumpsters. Including desert dust and bullet holes.
We are aware of the Israeli junk dealer. He sells his garbage over here too.
It's not all junk, my
hp 6920B came from one of those dealers about 10 years ago, price was £20 + shipping which was around 3x the asking price. They seem to be a lot more than that these days.
I did give up trying to buy the
hp 5321B they have, wouldn't take a sensible offer so feck it I bought one from the US instead.
David
There's another way if you change how you look at the problem from a partitioning perspective. Smaller the partition, the quicker consistency can be resolved as a rule to the point you can have transactional consistency. The trick is to redefine the partition size based on the operation in question and the participants rather than the current lazy arse cheap thinking which is have one partition with all participants in it. This is complicated and specific to every single problem domain and not generic which is why you don't see it.
But that is what Usenet is -- the hard consistency is only enforced in small domains, which produce atoms (RFC 5536 messages), who once committed are possible to refer to from other atoms, and you can create a thread of messages that constitute a discussion. This thread construct is possible to build in an asynchronous way, and it will never be perfect, but it will approach completeness with increasing age of atoms.
Yes, that is the happy puppy view of distributed systems consistency, but there is something to it.
That’s ok until someone doesn’t get an ancestor message. Which is where the consistency guarantee fails.
Im working on something in this space at the moment. Targeted at contract resolution.
I'm not aware usenet had any consistency guarantees. The Byzantine Generals problem limits the ability to have distributed consistency.
Where there is a distributed domain with one "brain" controller, then the domain is partitioned, each partition will (re)create a domain controller. When the partitions reconnect, one controller has to commit seppuku. The "split brain" problem is deciding which one and then ensuring consistency between the two partitions' data - and it a very difficult problem to solve.
The networking equivalent is ensuring a token ring has one and only one token circulating.
So, are you expecting to have 2/3/4/x phase commit protocols
There's another way if you change how you look at the problem from a partitioning perspective. Smaller the partition, the quicker consistency can be resolved as a rule to the point you can have transactional consistency. The trick is to redefine the partition size based on the operation in question and the participants rather than the current lazy arse cheap thinking which is have one partition with all participants in it. This is complicated and specific to every single problem domain and not generic which is why you don't see it.
But that is what Usenet is -- the hard consistency is only enforced in small domains, which produce atoms (RFC 5536 messages), who once committed are possible to refer to from other atoms, and you can create a thread of messages that constitute a discussion. This thread construct is possible to build in an asynchronous way, and it will never be perfect, but it will approach completeness with increasing age of atoms.
Yes, that is the happy puppy view of distributed systems consistency, but there is something to it.
That’s ok until someone doesn’t get an ancestor message. Which is where the consistency guarantee fails.
Im working on something in this space at the moment. Targeted at contract resolution.
[Zips up flak jacket. Puts on helmet with a flourish and an insouciant raise of the eyebrow.]
Have you considered Blockchain?
[Runs...]