If that's what you said I'm afraid it wasn't obvious to me. I took "Systemd causes log corruption where sane alternatives do not have such issues" at face value.

I don't know anything about journald's format, but if it suffers 'extensive' damage under challenging conditions that syslog handles fine then it's not an appropriate binary format for this job. That's not a fault of binary (which can be as robust as you want), that's a fault of the wrong kind of binary.

I make no defence of systemd, I only respond to implications that syslog doesn't have its own problems with lost and corrupt messages.

Good engineering would be to do both, not assume it's to be one or the other. This thread seems to be derived from 'Lennart said if there's corruption we rotate', I didn't see anything factual about the frequency and circumstances of corruption compared between different logging systems. It is a fact in my experience that all systems experience it occasionally.

You may know better than I do about systemd, which I don't currently use.

For Linux systems where the power is cycled often without warning, I have to use other kinds of logging for some things because syslog is too unreliable.

I have and I wasn't happy about it when using Chrome for something that should have remained private to the application's users.

I tried every combination of command-line options, including undocumented ones, to turn off reporting to Google, including the options that are for this purpose, and there was still a trickle of reporting things that I didn't want reported.

But that was a few years ago. Maybe Chrome is more privacy respecting now :-)

I don't mind that it talks to Google by default, after all there are some good services if you like them, and phishing protection (for example) is a good thing.
But I was surprised and disappointed that using all the options to turn off reporting didn't turn it all off.

It could be OpenGL.

I have a Linux laptop where X crashes killing everything, and the system effectively locks up when visiting any page that launches WebGL in _either_ Chrome or Firefox.

That's with the Nvidia driver too.

Strangely, other OpenGL applications don't cause any problems, they even work. Only Chrome and Firefox.

Imho until that sort of thing is fixed I don't consider WebGL is safe to use on a public web page. I reported it, but there didn't seem to be much interest in fixing it. "Oh it works for most people".

Utterly false. The idea that syslog doesn't have corruption is false. I have seen syslog corruption many times. Whether it's truncated lines, merged lines because half of an old truncated line has a new message appended, blocks of 4k zero bytes, or single bit or single character errors.

In particular if a syslog file is truncated mid-line by either disk full, system failure, filesystem bug or drive bug, the best thing syslog could do when it resumes after boot would be to rotate log files at that point, instead of appending to the truncated file.

These are quite rare, but not so rare that I haven't seen them maybe 50 times in 20 years in Linux syslog files.

I have no opinion on systemd particularly, but with regard to this single thing of rotating logs on detecting corruption, instead of attempting to patch the corruption or continue appending to the file, I think the right decision was made, from the perspective of an admin wanting the best available information after a problem.

This is not true, the faster Javascript environments use type analysis to optimise away hash table lookups.

That's good advice.

But there are plenty of things on most networks that aren't critical servers or devices you have the luxury to control and plan.

If you regard security patches as essential only for those things, you're doing defense in depth wrong.

Heartbleed affected clients too, and several things that aren't internet facing services.

Like when Ubuntu Server 13.04 didn't get a fix for Heartbleed because they discontinued support after 1 year despite the criticality of the bug and the servers seeing considerable use? All the official replies were "it's your own fault" and "change distro version immediately". Which you often can't do quickly. No users really expected 12.04, 12.10, 13.10 and 14.04 to get the fix while 13.04 in the middle was left out - except people who read the really, really fine print and took it seriously. Shipping the security fix would have been trivial and saved a lot of people a lot of work; they just refused on principle.

It was probably the first time many users found out Canonical had changed the support duration (that's why 12.10 got the fix).

Thanks! But too late. That machine died this time last year, after 6 years of excellent service. I moved on to new hardware.

Hopefully the xorg.conf is useful to someone else.

I've just looked up what people are saying about DebugWait, and I see the font corruption - that's just one of the types of corruption I saw!
But perhaps that was the only kind left by the time my laptop died.

Just a note to others, that DebugWait doesn't fix the font corruption for everyone according to reports. But, it's reported as fixed by the time of the kernel in Ubuntu 13.04 according to https://bugs.launchpad.net/ubu...

I stand by my view that Intel GPU support never quite reached "excellent" because of various long term glitches, although I'd give it a "pretty good" and still recommend Intel GPUs (as long as you don't get the PowerVR ones - very annoying that was, that surprise wrecked a job I was on). Judging by the immense number of kernel patches consistently over years, it has received a lot of support, and in most ways worked well.

Getting slightly back on topic with nVidia: Another laptop I've used has an nVidia GPU, and that's been much, much worse under Ubuntu throughout its life, than the laptop with Intel GPU. Some people say nVidia's good for them with Linux, but not this laptop. Have tried all available drivers, Nouveau, nVidia, nVidia's newer versions etc. Nothing works well, Unity3d always renders ("chugs") about 2-3 frames per second when it animates anything, which is barely usable, the GPU temperature gets very hot when it does the slightest things, and visiting any WebGL page in Firefox instantly crashes X with a segmentation fault due to a bug in OpenGL somewhere, requiring a power cycle to recover properly. So I'd still rate nVidia poorer than Intel in my personal experience of Linux on laptops :)

Except for the bugs. I used Linux, including tracking the latest kernels, for over 6 years with my last laptop having an Intel 915GM.

Every version of the kernel during that time rendered occasional display glitches of one sort or another, such as a line or spray of random pixels every few weeks. Rare but not bug free.

And that's just using a terminal window. It couldn't even blit or render text with 100% reliability...

I investigated one of those bugs and it was a genuine bug in the kernel's tracking of cache flushes and command queuing.
In the process I found more bugs than I cared to count in the modesetting code.

Considering the number of people working on the Intel drivers and the time span (6 years) that was really surprising, but that's how it was.

Nonetheless, statically linking with LGPL libraries in the form of uClibc is _extremely_ common in commercial devices running uClinux. Without providing any way to relink. Forbidden, but ignored.

As the AC implies, that's not interference from bad or unshielded electronics in the mobile (or it shouldn't be).

An ideal mobile transmits only what it's supposed to, on the correct RF channels to communicate, and nothing else.
Like all devices there will be other emissions, but let's assume it's very well made and effectively perfect.

The sound on the speakers is because the speaker circuit is effectively an RF receiver, converting those high frequencies to audio. They actually demodulate the signal - unintentionally. See http://en.wikipedia.org/wiki/Electromagnetic_interference#RF_immunity_and_testing

If the speaker circuit is made well enough, it won't do this.

I had exactly the same thing happen (audio stopped playing until reboot, so phone ring was silent) on Nokia Symbian S60 phones years ago.

It's crappy but it's not exclusive to Windows phones.

We stagnate unless we choose not to. You don't *have* to become a stereotypical angry old conservative. That's up to you. I choose not to.

Because Perl switched to a better hash function _and_ randomised it ages ago.

Having looked at many different fast hashing functions, I'm amazed at how many in the vulnerability report are still using the ancient multiply-by-small-constant and xor/add. That sort of thing tends to need a prime hash table size and a slow 'mod' operation. We have better hash functions that work on a 2^n table sizes.