* Posts by Peter Gathercole

3127 posts • joined 15 Jun 2007

Baroness Dido Harding lifts the lid on the NHS's manual contact tracing performance: 'We contact them up to 10 times over a 36-hour period'

Peter Gathercole Silver badge


The count of the number of tests being done was a meaningless number collected initially to show the government had achieved a target.

The reason why it is meaningless is that it did not indicate the total number of prople tested, as some people, like NHS workers, will have been tested multiple times to make sure they remained free so they could continue to work safely.

I don't believe that the articles actually say the tests were being stopped.

Not testing everybody who has been traced does not make any sense, however. Especially as there is no guaranteed sick pay for the people they're telling to lose 2 weeks income. People who absolutely need the income will just ignore the warning unless they either show symptoms, or it becomes a criminal offense to not isolate.

Peter Gathercole Silver badge

10 times

But if they manage to contact them on the first attempt, one would hope that they would not do the remaining 9 attempts...

Finally, a wafer-thin server... Only a tiny little thin one. Oh all right. Just the one...

Peter Gathercole Silver badge

Even enterprise grade UPSs can have problems

Late 1980s. Large telecoms development company. Mainframe data centre just outside a small Wiltshire town supplied by overhead power cables. Enterprise grade UPS with diesel backup generators.

What we learned was that multiple power brown outs during a significant thunderstorm was sufficient to defeat this setup.

The problem was that each time the power grid browned out, the UPS would kick in, switching to battery and temporarily turning off the air conditioning, which would have been switched back on once the generators started. The problem was that the power resumed before the generators started, so the UPS switched back to mains, and shortly afterwards, the aircon came back on.

Until the next brown out a few minutes later, and the next, and the next. Over the course of about 2 hours, the batteries became depleted, as there was no time to recharge them after each brown out, and the temperature in the machine rooms began to rise as the air-con was off so much of the time.

Eventually, even though the UPS should have been able to keep the whole DC running, it was decided to turn off the mainframe and the development and test environments, and halt work for the rest of the day,

I'm not sure why, but there the manual switch to generator in this setup had been overlooked in the design, which would have been able to keep the data centre running had there been one. But this taught me was that even professionally designed, very expensive UPSs are not a guarantee of continual operation.

Faxing hell: The cops say they would very much like us to stop calling them all the time

Peter Gathercole Silver badge

Integrated fax systems were a surprise to some!

Back in the early 90's, and very late one evening when I was providing on-call support, I was on a call from a customer who had managed to do an "rm -r" (fortunately on a data filesystem rather than /) on one of their systems, but who had very sensibly just hit the power button, and was wanting to recover as much as they could.

I sat on the phone with them for a while, talking to them continuously, while I worked out in the background (by deliberately corrupting a filesystem on one of our test systems) how to scan the filesystem using icheck and fsdb to work out which inodes with zero link counts still contained the block list for the deleted files.

Once I had this sorted, I got the fax number for their office from our customer records, and sent the script to it from my desk using our fax server. I then said I was sending them the script, and they asked how I was sending it, and how long would I take. By that time the fax had left the queue, and I just asked them if they knew where the fax machine for the number I'd sent it to was, and that they would find the script there. I could never understand why they were so surprised when they looked and found it. Shows that even technical people did not fully appreciate the advantages of an integrated IT system.

Once they knew that they had the script, and that it worked (it set the link count in the inode to one, and then let fsck sort out re-linking the file into the lost+found directory), I left them to it, saying that they could page me again if they had any further problems, which they did not do.

I got into the office the next day to find that the customer had completed the procedure, and had all of the data that they could not live without back (although not all of the files). The credit for closing the call went to the start-of-day person who called them back to confirm the state of the call!

And did I get any thanks? No. Of course not. I actually doubt that any of the other on-call specialists in the centre has the same knowledge of the UNIX filesystem, the fax machine setup and the test systems to be able to do the same.

Repair store faces hefty legal bill after losing David and Goliath fight with Apple over replacement iPhone screens

Peter Gathercole Silver badge

Re: Worse than expected, but that's just a detail @AC

I'm not sure you fully understand the case.

The issue is that Apple are using trademark legislation to prevent what could well be genuine parts removed from broken or failed devices to be used for repair.

Their assertion is that the parts are actually counterfeit, and seem to have persuaded the Norwegian legal system this is the case.

They do have a case. There are counterfeit parts manufactured and sold as genuine. But there are also genuine parts available. It is difficult to tell on cursory examination.

As I understand it, there is also a grey area where damaged genuine parts have the damaged elements replaced, for example, a functional genuine screen with damaged glass has the glass replaced, probably with the same spec. glass that Apple use, so is an amalgam of a genuine and after-market part. Does that make it counterfeit or genuine? I'm not sure, but Apple assert that it is counterfeit.

Apple says that anything they didn't supply that contains any Apple identifying symbols must be counterfeit, something that is almost certainly not true. As a result, they stifle the supply of parts to just that of what they deign to supply at whatever price they want to sell at, and the real counterfeit parts (which they also want to ban but have difficulty at the current time).

Once they get this, they can control parts supply to make it uneconomical to repair their products. In a normal supply-demand economy, this should damage their brand, but it seems that the buying public are just so enamored by that logo that they continue to pay large sums for devices that may well break and become un-repairable long before the customer expects.

So you really didn't touch the settings at all, huh? Well, this print-out from my secret backup says otherwise

Peter Gathercole Silver badge

Re: May I recommend rsyslog?

Ah. Syslog vs. Rsyslog.

Yes, Rsyslog was more recent, but syslog, the tool Rsyslog was supposed to re-implement and possibly improve does indeed go back to the 1980s.

Peter Gathercole Silver badge

Keylogger login program

Basic student prank back in the late 1970s. Write a shell script to emulate the login screen on a UNIX system. to capture the user ID and password of the next user of the terminal.

Second level, write it in C so that it would not display the password when it was typed (the stty program was less advanced on Edition 6, and did not have -echo and echo, and if you used raw mode, it was difficult to process end-of-line correctly in shell).

Was a war of wits. When you walked up to a terminal, you pressed return several times to cycle through to the real login screen, then the programs would notice no username and would loop. Then use EOF to get getty to recycle, and the programs would trap EOF and loop round. Then press break (it was the default interrupt on Edition 6) and the programs started changing the interrupt character.

The best of the login screen key loggers that were written were really sophisticated towards the end of the 'war'.

Eventually, the sysadmins started threatening to ban students engaged in writing these programs, but not before finding the source code and examining how they were doing what they were doing. UNIX was new back then, and everybody was learning from each other!

For the price tag, this iPad Pro keyboard better damn well be Magic: It isn't... but it's not completely useless either

Peter Gathercole Silver badge

Re: Because ... it’ll just work : Nope

Been recapping my BBC Model B's power supplies recently. Still working once that is done.

Peter Gathercole Silver badge

Re: Because ... it’ll just work : Nope

I have an IBM Thinkpad T23 in daily use as a firewall, with a PCCard Gigabit Ethernet adapter, circa 2000. Runs 24x7 355 days a year. It replaced a T20 from about 1996 when I got FTTC, and the 100Mb/s built in Ethernet became a bottleneck.

ALGOL 60 at 60: The greatest computer language you've never used and grandaddy of the programming family tree

Peter Gathercole Silver badge

Re: .. never used .. ?

Strictly speaking, it was PL/1 (Pea El One). although the 1 was oflen written as an "I" as in the Roman Numeral. But I get a bit upset when someone pronounces it as Pea El Eye, which people are prone to do.

But yes, it tried to be all things to all people, a scientific language, a business language, a control language and in some of it's incarnations (like PL/C which I used when learning PL/1 as a formal language in 1978), a teaching language.

It had many unusual features. The one that I found most interesting were implied loops in I/O statements that allowed whole or even part arrays to be written out in a single PUT statement.

The other language I was formally taught was APL (literally A Programming Language) of which I used to say (somewhat repetitively) "It's all Greek to me!"

Neither of them helped me with my first job, which was as an RPG2 programmer! Thank goodness I had taught myself C while at University. And I had no problem teaching myself Pascal at my second job.

Danger zone! Brit research supercomputer ARCHER's login nodes exploited in cyber-attack, admins reset passwords and SSH keys

Peter Gathercole Silver badge


How about agent forwarding?

It's not you, it's Slack: Chat app falls down – and at such a very convenient moment

Peter Gathercole Silver badge

Re: Enter the matrix @Robert

You're not thinking far enough back.

Before the rise of Google, Yahoo, Hotmail et. al., email was very distributed. Major operations ran their own smtp server which, provided that TCP/IP and DNS MX records continued to operate, would allow messages to get through even when there was some disruption. And TCP/IP was originally intended to be resilient, and DNS is, by it's very nature, distributed.

This wasn't very helpful for home users, but did work well for companies, and hey! there wasn't the demand for email from joe public.

Sure, an organisation's own server might go down, but it's really not that difficult to have multiple mail exchangers for a mail domain, and many did. But even if one companies mail server broke, the rest of the email infrastructure around the internet didn't, and even undelivered mail would eventually get through if the service was restored, or generate a bounce message to the sender after a timeout if it wasn't.

We're now suffering from single large suppliers of services becoming single points of failure, outside of the users or customers control. The very intent of the distributed Internet is being undermined by the Google, Facebook et. al. large service providers, and even companies that do understand, are putting their eggs in the AWS and other cloud providers baskets (Slack runs on AWS, yes?)

It's all looking a lot like when companies used to do their batch processing at computer bureaus as it was in the '60s and '70s, but on a vastly larger and more pervasive basis.

Behold: The ghastly, preening, lesser-spotted Incredible Bullsh*tting Customer

Peter Gathercole Silver badge

Re: Beware enterprising users...

I've recanted this story before, but when I was in the support for a large multinational business system supplier, I was the co-ordinating specialist for a consorium of educational customers who were getting a reputation for not checking or testing any fixes we supplied to them. Unfortunately, they were important for PR reasons, which is why they had an allocated specialist as a primary contact point.

One of my jobs every week was to call my contact there and ask whether they had made any progress in apply any of the updates or fixes they'd been given.

One frustrating day, I put into the problem record my true feelings, something along the lines of "Sheesh, <Customer name> applying any fixes? Not a chance!". It was only mildly derogatory, but what I didn't realize was that not only did the customer have a technical advocate, they also had a relationship manager who allowed them to read the problem records....

I was duly hauled into my managers office with the relationship manager, and whilst my manager privately agreed with my sentiments, he had to be seen to be telling me off.

Unfortunately, a few months later, the then relationship manager moved into the support centre - as my manager! Fortunately, he was quite a decent guy, and we actually ended up with a good working relationship once we had cleared the air.

Square peg of modem won't fit into round hole of PC? I saw to it, bloke tells horrified mate

Peter Gathercole Silver badge

Re: *Hisses & wards against Evil*

There was a K6-III as well, and it held it's own very well against the Pentuim-III, although not all Socket 7 motherboards would work with them.

Wikipedia says that the K6-III+ tops out at 500MHz, but I was pretty certain that I had one rated for 550MHz, although it could have been a K6-2+

Peter Gathercole Silver badge

Re: heh

If it was an early Optima, you were supposed to take the bezel off, and out the drive behind it....

Peter Gathercole Silver badge

Re: Saws

I don't believe you! Mjölner should be well enough known.

Mind you, I'm not sure what place a vehicle mounted mortar system has in a toolbox.

Getting a pizza the action, AS/400 style

Peter Gathercole Silver badge

Re: "Hopefully he also added a bit of text along the lines"

The dangerous commands comment should, obviously, carry a caveat of if you are running commands that you do not know what they do as a privileged user, you should have your privilege revoked immediatly.

I don't know how Unix got dragged in here, but ever since Unix edition 2 or 3 in the 1970s, you've had the concept of ordinary users and privileged users, so there has been no excuse to do day-to-day user tasks with a privileged account.

Peter Gathercole Silver badge

Re: "Hopefully he also added a bit of text along the lines"

The mantra with many operating systems is "If it completes silently, it worked".

This has never seemed a problem to me, at least not until you get a command that fails silently.

A paper clip, a spool of phone wire and a recalcitrant RS-232 line: Going MacGyver in the wonderful world of hotel IT

Peter Gathercole Silver badge

Long RS232 cabes

I have two stories about long-run RS232 cables.

The first was when I worked at a UK educational establishment. Pretty much everything was jury-rigged there, because they did not want to pay professionals to do things like laying cables. Anyway, my PDP11 was having some cables run down the hall to the lecturers offices, where their newly bought BBC Micro's were to act as terminals (and, in fact, were also connected to the floor Econet). The cables were laid, and loosely tied to a convenient support. Everything worked fine. One of the other technicians then decided ti tidy it up, and proceeded to tack it to whatever was available using staple gun instead of wire clips....

You can guess. Staple straight through the cable, shorting pins 2 and 3 (as we later found out). No immediate problems, because the BBC micro was not being used. But the PDP11 started getting slower and slower, and after about 4 minutes or so, crashed with a I/O buffer overrun error. Rebooted, and the system was fine for four minutes, and then crashed again.

Finally realized that on the port with the damaged cable, the system was sending out the login banner, which was promptly read back in as input from the terminal (pins 2 and 3 shorted). This generated multiple new login banners, exponentially increasing the amount of traffic until the PDP11 (which was very good at handling character I/O normally), just gave up the ghost,

Once found, rather than laying a new 25 metre cable, the rather embarrassed technician cut the damaged bit of the cable out and spliced in (using screw terminal blocks, I believe) a couple of inches of new cable,

The second story is from a factory floor, where a terminal was in the middle with the cable, the maximum length permitted by the RS232 standard, was run through the roof. Periodically, a couple of times a day, the computer it was attached to reported TTY Hog messages, and promptly shut the terminal down. I was giving remote support to a VAR, and after several days getting them running diagnostics on the port and checking the cable for damage or shorts, I asked where the cable was routed. They said that they had run it down the existing cable runs with everything else. After a few seconds thinking, I asked what else was fun down the runs. "Oh", they said. "Pretty much everything". Apparently, this included power for the electric motors that ran large industrial hoists for moving things around the factory floor. "OK", I said, "Is there any chance that the ports shutting down happened at the same time that these hoists were operating,,,,".

Turns out the motors were very dirty, and drew a lot of current. The rapid current spikes generated by the motors was being picked up as interference on the RS232 cable, which was not good quality shielded cable. It would have worked fine in an office environment, but in an industrial environment, it was not up to the task. The system tried it's best to make sense of the noise but failed, shutting down the port. Anyway, they actually relocated the terminal to a different part of the factory floor with a much shorter cable run and good quality cable, and I never heard from them again.

OK brainiacs, we've got an IT cold case for you: Fatal disk errors on an Amiga 4000 with 600MB external SCSI unless the clock app is... just so

Peter Gathercole Silver badge

Re: This can be easily explained

Virtual Address Spaces are much older than the Amiga and 68040s (I'm pretty certain that the 68030s and later had fully functional MMUs as part of the CPU).

I'm just trying to remember the computer architecture course I taught in the mid 1980s, where I talked about the first workable virtual memory system, which has to include a virtual address space. Ah. Atlas at the University of Manchester in 1959.

Peter Gathercole Silver badge

Re: the real answer is probably some shared memory corruption

IBM did this for more of their OSs than just AIX.

Peter Gathercole Silver badge

Re: My favourite timing bug


Interesting thing was we were running MDF, the Multiple Domain Facility, and one of the "domains" (read VMs for the younger readers here) was a full blown emulation of a 5EE3 telephone exchange!

Even though it was a really expensive mainframe, emulating one of the large telephone exchanges that AT&T Philips Telecommunications (APT) were selling was still cheaper than building and running one of the actual exchanges.

The systems all ran R&D Unix 5.2.5 or 5.2.6 (based on Amdahl UTS), which even though it was SVR2, had many SVR3 features before they made it into commercial releases, such as a paging virtual memory system, STREAMS and RFS.

Just after I left, the EE was ported to multiple Sun 3/280s and eventually SPARCs running across Ethernet, running R&D Unix 5.4, built on top of Sun OS 4.03.

UK government puts IR35 tax reforms on hold for a year in wake of coronavirus crisis

Peter Gathercole Silver badge

Re: One-nation Barmy @JamesPond @Mike 137

I agree that if the spouse actually does meaningful work a salary should be paid, but it's not uncommon for the spouse to be given a salary salary just below the NI LEL for doing nothing, so it attracts no tax or NI. That is not a huge amount of money, but there is no tax or NI paid on this money at all, reducing the tax take by HMRC.

I know this happens because when I ran my own company, the accountant I was using complained at me bitterly because I didn't.

Firefox to burn FTP out of its browser, starting slowly in version 77 due in April

Peter Gathercole Silver badge

Re: "FTP is an insecure protocol and there are no reasons to prefer it over HTTPS"

Goodness, do people still use Kermit? It's main use, I seem to remember, was as a fie transfer tool on things that only appeared to use a CLI terminal connection. I mean, I know that it worked over a network, but there were much better tools.

The last time I used it was to transfer files from a DEC mini to a BBC Microcomputer.

Next you'll suggest people use xmodem!

Peter Gathercole Silver badge

Re: "FTP is an insecure protocol and there are no reasons to prefer it over HTTPS"

And on top of this, many FTP implementations include a chroot jail actually in the server.

And even it it isn't, it's been very common practice (for about 30 years or so) to set up a chrooted environment for ftp explicitly.

The exceptionally low overheads of ftp have often kept it a as an option in bandwidth constrained environments. But it's time will end eventually.

Butterfly defect stripped from MacBook Pros, Airs by Q2 2020, reckons Apple analyst

Peter Gathercole Silver badge

Re: I liked the butterfly keyboard.. @Dwarf

The VT100 with the Advanced Video Option was, I believe, the first terminal to implement smooth scroll. The AVO added quite a bit to the already considerable cost of a basic VT100 terminal.

Most VT100 compatible terminals actually implemented a VT102, which was a cost-reduced VT100 with AVO built in, but did not have the expansion slot. The VT102 was the most popular of the VT100 family, although externally almost identical to a VT100.

In the UK, the most commonly bought VT100 compatible terminal was the PT100 made by Plessey, which also looked very similar.

I actually found the VT220 (LK201) keyboard a much nicer layout than the VT100, but I was not that keen on the softer feel of the keys over the VT10x terminals, and shortly afterwards I came across IBM Model F and Model M keyboards, with the Model M being the best compromise of layout and feel IMHO.

Post Office burned £100m in UK taxpayer cash on Horizon IT scandal legal fees, MPs told

Peter Gathercole Silver badge

Re: Disgusted of tonbridge wells @Boris

Completely agree with your sentiments, but...

These Subpostmasters were, in the main, not tried and found guilty. Most of them caved in to the threats, and pleaded guilty, even though they knew they were not.

Thus overturning such convictions is very much harder, because in any appeal, the Post Office's barristers will ask why, if they knew that they were innocent, did they plead guilty.

There needs to be a full, no barriers, investigation, with a promise that wherever possible, affected people will be fully repaid all of the money they lost with interest, and criminal convictions overturned, such that they would be left as they would have been if this hadn't happened.

The irony is that any damages would be paid either by the Post Office, or as a scheme funded by the Government. Either way, the tax payer would probably end up with the bill.

Peter Gathercole Silver badge

Re: "That doesn't make sense" ...

A fudged click and drag probably indicates either that you're doing to much when logged in as a privileged account, or that the permissions on the files were too lax.

Chips that pass in the night: How risky is RISC-V to Arm, Intel and the others? Very

Peter Gathercole Silver badge

Re: The trade war changed everything

I'm not sure how much the ARM decision is important, at least in the short term.

ARM are not the producers of ANY chips (at least not in production quantities). ARM chips in Chinese products all come from other companies, and I'm sure that some of these will be influenced by the US trade restrictions, but I'm also sure that some of them aren't.

So there is probably still a route to getting ARM processors fabbed outside of China for products.

But I know that there are fab's in China. Bearing in mind that it is probably not easy to rescind an ARM development license (and also taking into account China's track record of abiding by rest-of-world patent law), I would expect that Chinese ARM and other microcontrollers will still be available.

For example, the Kirin range of processors are a Chinese design fab'd by TMSC.

Peter Gathercole Silver badge

Re: Installed base

I'm not sure that the x86 instruction set is in any way similar to the VAX instruction set.

the 8086 is actually a linear development of the 8085 processor (itself a development of the 8080 and 8008) with some of the *Ziliog* Z80 concepts added.

The VAX instruction set was a 32 bit development of the 16 bit PDP11, which was a very regular and orthogonal set, so much so that many of the register and memory addressing modes were implemented using the same mechanisms (for example, the Program Counter was manipulated using the same instructions such as auto-decrement as the other registers).

Use of the registers were more generalized in the 8086 than the 8085, but they were still more specific than the VAX.

Any resemblance is at the conceptual level.

Peter Gathercole Silver badge

Re: Installed base

Just to remind you, ARM-1 and 2, SPARC and MIPS RISC processors were available before POWER (the original designation for the RIOS chipset, Performance Optimization With Enhanced RISC). IBM also had the ROMP (801) processor before POWER as well.

IBM POWER was only marginally a RISC processor, as it's instruction set had a lot more instructions and addressing modes than other more traditional Reduced Instruction Set Computer implementations (and quite a few CISC processors). In addition it's initial implementation was a 5 or 7 chip set (as was the multiple chip HP PRISM processor) rather than a microprocessor.

And in fact the Intel x86 processors from probably about the 486 have embraced RISC, with the processor being a micro-coded RISC engine that executes x86 instructions on the surface, but actually JIT compiles them into micro-instructions.

Meltdown The Sequel strikes Intel chips – and full mitigation against data-meddling LVI flaw will slash performance

Peter Gathercole Silver badge

Re: One day, not to far in the future, @Bronek

I would guess that there are several things that could cause the initial flurry of disk activity. It may be that your system is set to do a scan of some sort every time it starts. Alternatively, it may be pulling into memory the scanning engine and the current virus definitions, as having this resident in memory would be a big time saver.

I don't actually run any Windows systems now (at least, not on a frequent basis, there is one that gets turned on very infrequently, so I can't relate any personal experience.

Peter Gathercole Silver badge

Re: If these exploits carry one

I can see that you're not that familiar with process and thread dispatch on current multi-core processors.

There are many long answers, but I'll try to give you a couple of short ones.

1. Lots of small processors are good executing lots of small threads/processes. They're bad at executing small numbers of large processes.

2. Accessing memory becomes exponentially more of a bottleneck as the processor count increases unless you implement a NUMA model.

3. If you choose a NUMA model, maintaining consistent performance becomes more difficult as memory closeness to the core executing the code becomes relevant.

4. The more processors you have, the more difficult it becomes to keep your memory caches consistent, which in itself can allow sideband memory attacks.

5. If using a monolithic kernel, locking contention on kernel memory structures by processes running on different processors can cause performance issues.

The large multi-core systems that can be bought now try to eliminate some of these problems by using static processor allocation to VMs, allowing the system to allocate memory for VMs based on affinity to the cores in a NUMA implementation, and to reduce the number of processors contending for kernel locks.

I'm sure there are lots of other reasons, but these came off the top of my head.

Peter Gathercole Silver badge

Re: Dumb question time

The problem with fixing it in the compiler/assembler is that you have to trust that all the code that runs on your systems has been compiled on a patched development environment.

Even if a system has a fully patched development environment on it, if you take a binary compiled anywhere else, you have to trust that they have a fully patched environment. And you also have to have all of your libraries recompiled.

You can bet that any nefarious player who is trying to hack your machines by dropping pre-compiled binaries on you system will *NOT* use a patched environment, so you have to be super scrupulous about where the programs you use come from.

Peter Gathercole Silver badge

Re: One day, not to far in the future,

For a lot of current machines, especially those on the desktop, the anti-virus protections are the biggest culprits for the perceived poor performance. If everything pulled through the network ports, or off the disk is scanned before it is used, this makes almost everything you do take much longer.

This is one of the reasons why putting Linux, especially one of the lightweight ones on a struggling Windows system makes it feel much faster.

Of course, you could turn off the anti-virus...

OK. It's Windows. Better not.

Open-source, cross-platform and people seem to like it: PowerShell 7 has landed

Peter Gathercole Silver badge

Re: Bash gets the extend, embrace, extinguish treatment @phuzz

Agreed about systemd. I'm still debating whether to shift to Devuan. I know I do not like trying to fix a systemnd system that does not boot correctly.

On the subject of PowerShell, of course Microsoft are not directly pushing it. They don't have to.

Very few organizations deploying Linux are pure. Most run MS operating systems as their primary environment, with an enclave of Linux. In this type of environment, the Windows admin teams will lobby to have PowerShell on the Linux systems, not Microsoft, because they will point out to the managers how convienient it is to have a single management system, and <how much easier it would be and cheaper</i> if they could manage parts of the Linux environment using the same infrastructure as the rest of the environment.

And because the managers are conditioned by the bottom line on their budget, they will believe the Windows admin's over the objections of the Linux admin team, who now see their jobs at risk.

I've seen the loss of influence of the *IX admins, and the corresponding rise in Windows almost everywhere I've worked over the last 20 years, and it's not going to stop now.

Peter Gathercole Silver badge

Re: Bash gets the extend, embrace, extinguish treatment @phuzz

What has made *IX administration with Bourne and derived shells so long lasting is the ubiquity of it.

I have found that if you keep to Bourne shell syntax, for all it's limitations, the resultant scripts will just work in ksh, bash, ash et. al. (well, there are some differences, like the order of pipeline construction and the inheritance of file descriptors, but these are corner cases, albeit not common).

This is partly your "whatever you're most comfortable with" point, but it is also something you've not touched on. It's standard and available almost everywhere, and it's not just what I'm comfortable with, it's what every *IX administrator is comfortable with.

If we get to the point where, for example, PowerShell, ipython, Rush or psh is mandated for administering a system, then firstly, tradditional *IX admins have a learning curve ahead of them that they would not have with a posixy shell, and secondly, how stable/reliable/maintainable is the system (I'll pick Rush as an example here which almost got abandoned).

On top of that, you've got a very strong 'not invented here' streak in the community, especially when Microsoft are associated with it.

Peter Gathercole Silver badge

Re: Bash gets the extend, embrace, extinguish treatment

Whilst I agree with your general sentiments, there are things that PowerShell can do that ksh, bash and all of the other derived shells just can't.

Unix shells are great at the "stream of bytes arranged as lines" way of passing data, and when you are in a CLI environment, where people interacted with the systems through line-by-line interfaces (as opposed to form or even GUI based admin methods, this works well. As Linux is a derivative of UNIX (albeit a non-linear re-implementation), and most things have command-with-arguments or file based administration methods, things work great. Even where there are GUIs, more often than not they are grafted on top of the shell commands that actually do the work, which you could just have easily run from a shell script.

But things are changing. More and more, settings are stored in XML or object based storage, and unless you have a command that you cal call from the shell to manipulate these object, shell does not hack it anymore.

Some time ago, I tried to control some KDE processes (specifically knotes) that talked via kdbus or dbus or somesuch, and there were some objects that could be returned which did not map conveniently into something that a shell process (even using awk to help) could cope with, because the object-to-object mapping was very difficult to represent.

The more OS administration relies on objects that can no longer be represented in stanza or delimited files (and I include XML and related files in the 'difficulr' category), the less likely it is that bash et. al. will be able to hack it.

Of course, you could say "what's wrong with stanza or delimited files", but that is a completely different discussion, but basically as the systems get more complicated, the associations between different subsystems and objects just get too complicated to represent in flat files.

I don't like PowerShell becoming the default shell for administering *IX systems, but there is a need for something with more that flat file manipulation, and unfortunately us UNIX and Linux admins have been able to just about hack it using script gymnastics in Posix type shells and related commands up until now, so nothing more capable caught on in our space. We're as much to blame as Microsoft (I'm sure there are object based shells, but I can't name one off the top of my head, which shows how well they've penetrated the *IX space).

I feel really old. As a 40+ year veteran of UNIX and related systems, I am used to the traditional ways of doing things, and all of the PowerShell, systemd, Object and database based configuration and software communication busses just make me think that I'm past the point of being able to move on in to the future in the IT industry.

But I do wonder whether all this new complexity is actually worth it in the long run. Soon you will need an AI just to be able to administer some of these complex systems, and human beings just won't be up to the task.

Try to fix a broken system? You're having a laugh!

Brit MPs, US senators ramp up pressure on UK.gov to switch off that green-light for Huawei 5G gear

Peter Gathercole Silver badge

Re: Why is it a risk? @DiViDeD

Ah, Sheffield again.

A Type 42 Destroyer which was not sunk by the Exocet that hit it and passed right through without exploding because it was not armored, but by the fire that the rocket motor caused in the flamable wiring insulation, which led to a complete power loss preventing the crew from fighting the fire.

OK, it was a French missile, but in reality it could have been pretty much anything.

You should really have quoted the Dassault Super Étendard or the Italian Aermacchi MB-339, but then they also used American Douglas A4 Skyhawk aircraft and even British Tigercat and Blowpipe missiles.

Their navy also had two British designed (one of which was built in Britain) Type 42 destroyers like Sheffield, several ex-American cruisers, destroyers and submarines, and even an ex-British aircraft carrier!

Countries without significant native arms industries tend to by their weapons from anybody who is prepared to sell to them!

Running on Intel? If you want security, disable hyper-threading, says Linux kernel maintainer

Peter Gathercole Silver badge

Re: Updating Firmware isn't easy

The boot loader for PDP11s was short enough so that you could key in the 10 or so instructions from the front panel to initiate the load of the first stage bootstrap. From the V6 "Setting up UNIX" document on Tuhs:

Once the UNIX `binary' disk is obtained, the system is booted by keying in and executing one of the following programs at 100000. These programs correspond to the DEC bulk ROMs for disks, since they read in and execute block 0 at location 0.

RK05 RP03 RP04

012700 012700 (to be added)

177414 176726

005040 005040

005040 005040

010040 005040

012740 010040

000005 012740

105710 000005

002376 105710

005007 002376

Many early machines actually had the boot loader on paper tape that would be run through a reader on the console.

Sorry about the line spacing, the < pre > marker looks like it does not preserve line spacing.

Computer, deactivate self-destruct system requirement, says Sonos... were it on a starship in space, and not a smart-speaker slinger

Peter Gathercole Silver badge

Re: Surprised Sonos survive? @Jason

Well, if you were an audiophile, you definitely wound not buy Sonos equipment, except maybe for lift or background music.

Is that a typo? Oh, it's not a typo. Ampere really is touting an 80-core 64-bit 7nm Arm server processor dubbed Altra

Peter Gathercole Silver badge

No point @Stuart

RISC-OS can only use a single core. There is no MP code in it, and adding it would require a major rewrite. In addition, it is only 32-bit, and I'm not 100% certain that the 64-bit ARM processors still have 32-bit instruction modes.

When I first learned the in-depth technical aspects about MP for Amdahl UTS (a UNIX SVR2 port initially) running on a mainframe, we were told about just how much of the kernel had to be changed to add spin-locks on all of the kernel structures, and the number of man-years that it took.

RISC-OS is more simple, but the effort would still be significant, especially as the process model for RISC-OS does not use the hardware address protection that the ARM provides, so the lack of process address space separation would also be a major problem.

US Homeland Security mistakenly seizes British ad agency's website in prostitution probe gone wrong

Peter Gathercole Silver badge

Re: US Homeland Security

I get somewhat worried by people who have decided that .com is too long, and want to register in .co

What is even more worrying is when I remind them that .co is the country domain for Colombia, and they are not too concerned, even when I remind them about their ongoing issues with marching powder.

Not really sure I want to have financial transactions with a website in .co!

We regret to inform you there are severe delays on the token ring due to IT nerds blasting each other to bloody chunks

Peter Gathercole Silver badge

Re: Token Ring in the early '90s?

And to cap it all, 4Mb/s Token Ring! 16Mb/s was available in the '80s.

Token Ring had a traffic protocol whereby the packet would be sent by the sender, and received by the target, which would then set a flag and re-transmit the packet, still full. It would then flow around the ring until it arrived at the sending station, which would then 'empty' it and re-transmit the empty token to the next station on the ring. The sender of the original packet, which had just emptied it, was not allowed to fill it. In this way, it was made impossible for a small number of systems to monopolize the ring. All stations would get a crack at sending, as eventually the token would arrive 'empty' at every station.

Unfortunately, I believe that some of the non-IBM network drivers for Token Ring would not honor the rule about not immediately re-filling the token, and this allowed other stations on the ring to get 'locked out'.

Ethernet, both 10base5 and 10base2 would also seriously suffer under conditions of high congestion (because of CSMA/CD), and it was not until the advent of the 10baseT switches (as opposed to a hub, which acted much more like a cable than a slotted bus), that congestion problems began to go away.

It was always said, when I learned about networks, that Token Ring had higher overhead than Ethernet when the network was less busy, but coped better with high network loads.

I believe that the very last incarnation of Token Ring, using Madge intelligent CAUs and RJ45 Structured Cabling rather than MAUs, actually allowed more than one token to be passing around the network at any time (the number depended on the number of stations on the network), with some form of buffering to copy with stations with different speeds. This made it more like the 'token bus' that an earlier poster mentioned. But by that time, 100baseT and faster was becoming faster and cheaper.

I ran an IBM Call Center using Token Ring for a number of years, and I have a number of tales I could tell about how easy it was to break a Token Ring network!

Wi-Fi of more than a billion PCs, phones, gadgets can be snooped on. But you're using HTTPS, SSH, VPNs... right?

Peter Gathercole Silver badge

Re: "MitM attacks on unencrypted network traffic do happen"

Even your hidden ESSIDs for WiFi networks are visible, they just don't broadcast their name.

I use Kismet on Linux to get a picture of the WFii networks around me, which shows a very alarming situation where I stay when I'm working away from home. There are over 20 networks within range, over both 2.4 and 5GHz bands. Causes significant congestion and connectivity problems when everybody is streaming media in an evening.

WiFi just doesn't appear to be that suitable for large blocks of flats.

RIP Katherine Johnson: The extraordinary NASA mathematician astronauts trusted over computers

Peter Gathercole Silver badge

Re: Amazing woman

My wife is a real technophobe, and when I suggested that we watch Hidden Figures one evening, and I described it as a film about women involved in the background of the US space program, she heard "space program" and "mathematicians" and said that she didn't want to watch it.

We didn't find anything else she wanted, so I put it on anyway, and by about a quarter of the way through, she had changed her mind, and by the end, she was completely engrossed. So even if you don't think that you want to watch a film about the space program, watch this one. I promise it is worth seeing.

There was some artistic license apparently, but I think it was a fitting tribute to these women of colour who had everything stacked against them, and still managed to make a difference.

Your mission, should you choose to accept it, is to save data from a computer that should have died aeons ago

Peter Gathercole Silver badge

Re: Serial Taps

DECs RS232 setup was pretty conventional. IBM 6150 / RS6000 was less so (especially when you consider the unusual 10 or 12 pin connectors that IBM used in the built-in ports). If I remember correctly, IBM used a combination of DCD/DTR/DSR handshaking, but both systems would be DTE (Data Terminal Equipment) devices, so there would have to be some sort of null-modem involved. The secret was connecting DTR on one end to DCD as well as DSR at the other, probably both ways.

I never got to play at a hardware level with a DECServers, but I would expect that they were not dissimilar to the serial ports on KL11 and DL11 serial cards in PDP-11s, which I did a lot of work with. Should not have been too difficult to get working, even using off-the-shelf cables and null modems.

Configuration wise, it depended which way you were going, or whether you wanted both ends to initiate a connection. At the IBM end, you would have had to set the terminal line up as 'pshare'. The other end? Well, that would depend on what OS you were using on the VAX.

Peter Gathercole Silver badge

Re: Got one on the go..

Get some 10base2 to 10baseT bridges. I'm sure that you can still find them on Ebay, although only being 10Mb/s, you're probably going to have to plug this through a switch that can still talk down to 10baseT.

Heck, I think I may still have one sitting in a crate at home.

Bloody hell, someone is trying to sell an AUI to 10base2 transceiver there! Now that is seriously obsolete.

Peter Gathercole Silver badge

Re: Serial Taps

The number of times I've used RS232 as a communication of last resort.

I remember using a hex editor to capture a CP/M 80 program using a BBC micro as a data capture device, and then writing a program to turn it back into a binary so it could be run on the Torch Z80 second processor which ran a CP/M 80 rewrite called CPN. (CP/M machines were notorious for not having a standard disk format that allowed data transfer between machines).

I frequently used BEEBs as data capture devices, and wrote my own DEC VT52 emulator, and Tek 4010 emulator (this was before Termulator was available). I used to capture the graphic instructions for sessions from the MTS mainframe onto the BEEB so that I could re-display it while not connected to the mainframe.

I later became known as the terminal king at several places I worked, as I could nearly always get a terminal working on a serial line, and home-made interposers made using DB25 connectors held together with long bolts and nuts, with soldered wires between the connectors. Together with a Tektronic logic analyser, I was unbeatable! DSR, DTR, DCD, CSR, CTR, chassis ground and signal ground, XON/XOFF, I pretty much came across it all.

I was also the UUCP/BNU SME wherever I worked, whenever they wanted to do mail exchange (pre-TCP/IP) I was almost always involved somewhere along the way. Add to that Termcap/Terminfo descriptor file writing and editing, and that rounded out my serial credentials.

Printers, X.29 PADs, modems, terminal servers, reverse telnet for calling out through modems connected to the terminal servers, I did it all. The most strange thing I got involved in was a weigh-bridge that was connected to an IBM 6150 Unix system. That was a little strange.

All skills long dead, but I do still find it interesting that in these days of xterm being the (very poor, there was no one single xterm type) lingua-franca of terminal emulation, Linux still has a full-blown terminfo database and ncurses implementation, containing references to terminal types that were long obsolete before the turn of the century!

Don't use natwest.co.uk for online banking, Natwest bank tells baffled customer

Peter Gathercole Silver badge

I actually think it was a buy-out. Nat West got themselves into an awful mess and nearly went down in the late '90's, and were vulnerable to a hostile buy-out, and that is what RBS did.

I never understood how a smaller bank was able to buy a bigger one, even if the bigger one was in trouble, but such is commerce.


Biting the hand that feeds IT © 1998–2020