UNIX Socket FAQ

A forum for questions and answers about network programming on Linux and all other Unix-like systems

You are not logged in.

#51 2014-01-17 12:13 AM

i3839
Oddministrator
From: Amsterdam
Registered: 2003-06-07
Posts: 2,230

Re: Vacation time again...

The mind boggling part of relative velocity time dilation is that it's always the other that seems to be going slower. I was wondering about that before, now it all makes more sense.

Offline

#52 2014-07-16 12:40 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

I'm off again for 2 weeks starting this weekend...

Offline

#53 2014-12-03 04:07 AM

vic
Administrator
Registered: 2002-04-16
Posts: 85
Website

Re: Vacation time again...

UUUUhhhhhhgggggg....... I'm a zombie process. Just try to kill me. You can't!

I just got an email from somebody due to my involvement with the ancient UNIX Socket FAQ. I feel like a legit Living Dead coder now.

Anyway, I hope you are all well. My last three working days have been writing unit tests against AngularJS code (which I'm responsible for). Only two modules out of about twenty made me throw up my hands and give up, so yay me. I hope they'll be re-written soon. Also by me.

Muchos love to you all.

Offline

#54 2014-12-05 10:06 PM

i3839
Oddministrator
From: Amsterdam
Registered: 2003-06-07
Posts: 2,230

Re: Vacation time again...

Hey vic, good to see you around!

AngularJS actually looks interesting for my work too. At my work we make embedded systems,
but those are configurable through a web interface. I only work there for less than a year, so I
couldn't have prevented the big PHP mess we have now. We need to rewrite the thing, and I want
to push as much dynamic stuff to the client side as possible, so that we can use the same configuration
interface for the big systems as for the stand alone units which don't have much computing power.

I'm off to Australia for a month of vacation this Saturday, but that will
probably actually result in me having more time to check this forum.

Offline

#55 2014-12-08 11:33 AM

Nope
Administrator
From: Germany
Registered: 2004-01-24
Posts: 385
Website

Re: Vacation time again...

I am also still alive... :)

Offline

#56 2014-12-08 12:45 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

I think I am, as well, but it's too early to be sure...

Now, we just need Michael, Biologz, and Loco to show up!

Offline

#57 2015-03-20 05:48 PM

biologz
Administrator
From: Puking on the pavement
Registered: 2005-11-02
Posts: 396

Re: Vacation time again...

And Biologz is showing up :-)

I spend a few hours reading old posts and it remembered me all the good times I spent wit you guys here... I miss it sometimes, hehe, I miss computers too, sometimes...

How are you all? I can see that the forum is still active and that you admins are still, and even better, geniuses :-)

PS: man how many bullshits I posted here on this forum :-)


gethostbyintuition() is still a dream of mine

                                                 -- quoted from bash

Offline

#58 2015-03-20 08:20 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

Hey, good to see you around again!

Well, I wouldn't call the forum very "active" these days...  We get maybe one real new thread per month, if that...  I think most of the cool kids these days have moved onto other more general coding forums like Stack Overflow...  Or else not many people are writing low level sockets code at all these days...  We've always had something of a niche audience here, but it used to be we could at least count on an influx of newbies asking questions (or wanting us to do their homework for them) around the start of every new semester, but we hardly even get much of that anymore...

But, I'm doing ok...  Nothing much new, aside from the IPv6 Net connection that Comcast just sprung on us by surprise a little while back...  I can't wait for them to roll it out to everyone, and maybe we can finally be done with the limitations of IPv4 and the evil that it necessitated (NAT) once and for all!  I'd almost be fine with their near monopoly on Internet connectivity in the US if they actually ended up getting us all on IPv6 finally... ;-)

Offline

#59 2015-03-20 08:40 PM

biologz
Administrator
From: Puking on the pavement
Registered: 2005-11-02
Posts: 396

Re: Vacation time again...

Hehe

I'm glad to hear from you Rob :-)

Ipv6... Well, maybe a few years ago I would have been enthusiastic about it, but I must admit that right now, I feel so far away from those things... Plus I really lost a lot of my knowledges concerning everything about computers (though with some time spent on catching up I think it could come back...).

So it's not that active here... How are you living with it, I mean you and i3839? I remember when I used to hang around here, there was so many questions, it was cool, even if I couldn't answer to a lot of things...

What are you working on right now?

I always wondered, but never ask, do you participate in some opensource projects?


gethostbyintuition() is still a dream of mine

                                                 -- quoted from bash

Offline

#60 2015-03-21 03:38 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

I suppose it gives us a break, and more time to do other stuff than post here...  Hell, even the spam has slowed down and doesn't take much time to deal with these days! ;-)

I don't regularly work on open source projects, mainly just because I don't really have the time...  But, often I'll need to port something open source to QNX for work use, and I might end up submitting some patches due to that...

Right now, I've just been playing with multicasting, which I've rarely ever done anything with before...  I just finished off a set of library functions to deal with all the sockopts and such for joining groups and binding to interfaces and what not...  Now, I'm trying to think of where it would make sense to actually use them!  Most people wait until they have a need for something before coding up a solution, but all too often I create solutions in search of problems, just because it seems cool or fun... ;-)

Offline

#61 2015-03-22 01:42 AM

biologz
Administrator
From: Puking on the pavement
Registered: 2005-11-02
Posts: 396

Re: Vacation time again...

Ok Rob.

I really hope this forum will stay alive. I mean, just the level of the admins is a good reason (by the way I'm not counting myself lol, but I noticed I'm still an admin here...).


gethostbyintuition() is still a dream of mine

                                                 -- quoted from bash

Offline

#62 2015-06-29 11:30 AM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

Well, I'm going away again on July 1st, and will be gone a few weeks...  Other than spam, I doubt there will be much else to deal with anyway...

Offline

#63 2015-07-01 07:00 AM

i3839
Oddministrator
From: Amsterdam
Registered: 2003-06-07
Posts: 2,230

Re: Vacation time again...

Hi Rob, enjoy your vacation. I'm going away next week for one or two weeks, but I doubt anyone will notice.

The time of fora seems to be past, a new generation is growing up with those weird tiny portable computers and they do everything differently.

Offline

#64 2015-07-20 03:52 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

I'm back...  Doesn't look like I missed much...

I never had any use for cell phones so I'm one of the few people on the planet without one still...  Though, the fact that they ARE now "tiny portable computers" with Internet access anywhere has tempted me into thinking about getting one... ;-)

Offline

#65 2015-07-20 08:05 PM

i3839
Oddministrator
From: Amsterdam
Registered: 2003-06-07
Posts: 2,230

Re: Vacation time again...

I just came back too and deleted the spam already. Wasn't too bad.

I've got a mobile phone, but a very stupid one. A smartphone with decent screen is very handy when travelling, and that's the only reason I would get one. However, if battery life isn't at least a week then it's still fairly useless. And it has to be at least waterproof, otherwise it's a huge waste of money (I'm close to water too often).

Offline

#66 2015-12-17 08:39 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

I'm heading off for Xmas week this coming Saturday...  Not that we get much other than spam these days anyway...

Offline

#67 2015-12-19 07:34 PM

i3839
Oddministrator
From: Amsterdam
Registered: 2003-06-07
Posts: 2,230

Re: Vacation time again...

Have a nice time!

Offline

#68 2015-12-29 04:36 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

And, back again...  The place seems to have remained as dead as usual...

Offline

#69 2015-12-31 03:06 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

I think the spambots are devolving...  They used to post semi-coherent random babble, clearly cut-and-pasted from posts by real people from some other forum, but the last two I deleted were just plain gibberish scattered with links...  They must be using something like Markov chain text generation, or something even dumber than that, because it was really purely incomprehensible gibberish...  Oh well, at least it provides a bit of amusement when deleting them, anyway...

Offline

#70 2016-01-02 11:57 AM

i3839
Oddministrator
From: Amsterdam
Registered: 2003-06-07
Posts: 2,230

Re: Vacation time again...

The main point of those is that they're all unique, so can't by easily filtered by (e-mail) spam filters.

The time approaches to put the forum in read-only mode...

As a total side note, I used to like IPv6, but looking at it from a I-have-to-implement-this-shit point of view it looks more annoying to parse than IPv4. And the overhead is ridiculous: 40 bytes for basically no extra information?

The AMS-IX statistics say that 40% of packets are 64-128 bytes:

https://ams-ix.net/technical/statistics … stribution

Sure, in total bandwidth usage those 20 bytes are only 1% or 2% extra overhead, but if you basically send only lots of small packets, like we do at my work, then it's a step back from IPv4.

Offline

#71 2016-01-02 04:03 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

Yeah, it's extra overhead certainly...  For a while, I was with the folks who said they should've just gone with 64-bit addresses instead 128-bit...  But, over time I found the idea of the nearly infinite address space a wiser and wiser choice...  Being able to just hand out a /48 or /56 allocation to literally every single person on the planet (and then some!), and know that gives them all a shitload of separate subnets (64k or 256) if desired, as well as an insane amount of addresses within each (2^64), just seems like the best approach with the least hassles, and removes the need for end users to do utterly stupid bullshit like NAT to put all their machines on the Net...  Coupled with the autoconf stuff that self-generates the bottom 64 bits of an address, and all IPv6-capable routers sending out the top bits to all hosts via multicasting, and it becomes nearly transparent and hassle-free to end users...  For a while, when we were on Comcast and they added IPv6 support, some of our machines were just automatically using it without us doing anything or knowing anything about it...  I didn't realize my own machine was using it until some time after it was when I happened to do "ifconfig" and spotted the public IPv6 address assigned to the Ethernet interface...  That's just spiffy shit...  Sure, it's not that more magical than DHCP from an end-user's perspective, but the way it works is far cooler to me as a network coder...

One thing I think they got wrong with IPv6 was the printable address format...  I like going with hex instead of decimal, but why the fuck the colons?!  They should've stuck with periods to separate the bytes...  Or, if they wanted to use something clearly different from IPv4 addresses, commas or slashes or something...  There is a LOT of code out there that supports things like "<ip>:<port>" syntax, so the colons make that a real pain in the ass when going IPv6...  And, that klugey bullshit bracket syntax they've come up with the kluge around the issue is crap, too...

Offline

#72 2016-01-10 04:13 PM

i3839
Oddministrator
From: Amsterdam
Registered: 2003-06-07
Posts: 2,230

Re: Vacation time again...

I think it's just being lazy, 128 bit is plenty for anything, so it's an easy solution. But seriously, if you really need anything approaching 128 bits, a hop limit of 8 bits isn't enough.

The autoconf stuff and other mandatory features like multicast support are nice, but independent of packet format.

My problems with IPv6 packet format:

1) Unused Flow Label part. If you don't know what to do with it, just drop it, but don't add unused field to a new protocol.

2) The options within extension headers insanity. Why!? Why the layered approach which only makes everything more complicated for no good reason whatsoever. So an overly complex encoding to encode... nothing. Because there aren't really much useful options. But every IPv6 host has to support all the parsing and complexity anyway, just in case or something.

3) Order of extension headers and options isn't fixed. So routers have to parse all extensions, just in case.

4) What the hell were they thinking when deciding about fragmentation support?! Fragmentation is a pain in the ass and a security risk. Just get the hell rid of it for crying out loud. No, end hosts need to support it. Oh, but only for packets that reassemble up to 1500 bytes. Oh, and by the way, minimum MTU went up to 1280. So we're all supposed to jump through hoops because of 220 bytes? Really?

Offline

#73 2016-01-10 05:55 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

I thought the only extension header that routers need worry about is the hop-by-hop options, which must always be the very first extension header (if present at all)?  Unless it's doing some deeper packet inspection/filtering or something, in which case it might need to traverse the chain of extension headers...

I agree about the fragmentation...  One thing they did right was to remove it from routers' duties...  It's always as if the don't-frag bit is set for all packets...  That at least keeps things simpler for routing, and pushes the stupid frag duties to end hosts...  But, yeah, you're right that it's kind of silly to bother with it at all given the raised minimum MTU and maximum reassembly size limits... 

I don't really get the whole flow label thing, either...  It's apparently meant for speeding up routing by getting routers to cache all necessary processing based on (a hash of) source address and flow label, such that senders guarantee nothing has changed since the first packet with the same source and flow in all future packets with the same source and flow...  But, I have no idea if it's actually used in the real world at all...

Offline

#74 2016-01-11 08:50 PM

i3839
Oddministrator
From: Amsterdam
Registered: 2003-06-07
Posts: 2,230

Re: Vacation time again...

After reading RFC6437, the main purpose seems to be for load balancing without reordering packets (which they call the stateless use).

The Hop-by-hop extension header must indeed be first, but it contains one or more TLV-encoded options, so as far as parsing goes it's not easier for routers. "The only hop-by-hop options defined in this document are the Pad1 and PadN". So there are no mandatory useful hop-by-op options, but you're not allowed to ignore them either.

Routers must check for Routing Headers, which can be anywhere, so they have to be able to go through the whole list of extension headers. Granted, I thought they had to do it all the time, but they only have to do it when the packet is addressed to them. So in the end it makes no difference for code complexity, just performance. (And the only defined routing header type 0 is deprecated for security reasons already.)

Only one left: Destination Options Header. But for that one also only pad1 and padN are defined.

A full implementation of IPv6 includes implementation of the following extension headers:

           Hop-by-Hop Options
           Routing (Type 0)
           Fragment
           Destination Options
           Authentication
           Encapsulating Security Payload

Summary:

Only useful options are the security ones, the rest is non-existent, deprecated or stupid (fragmentation).
They should just drop everything except the encryption options and have a uniform encoding scheme for all options, be it TLV or not. Get rid of the onion, give me a pancake please.

Offline

#75 2016-01-11 09:26 PM

RobSeace
Administrator
From: Boston, MA
Registered: 2002-06-12
Posts: 3,827
Website

Re: Vacation time again...

There are a few other options defined in various other RFCs that can appear in hop-by-hop or destination option headers...

Offline

Board footer

Powered by FluxBB