Saturday, 29 October 2011

Stranger key on my keyboard!

Well, after ordering a new keyboard a while ago and having one of those strange double glazing keys on it, next time I ordered a keyboard I picked "linux" from the options.

I was expecting that maybe I would get a penguin or "super" or some such on the key in question.

Imagine my surprise when the keyboard arrived with the same strange key as before, but with caps lock and left control reversed, as well as escape and tilda reversed.

What is especially strange is the reversals are not just printing, but electrically as well. I thought (perhaps wrongly) that the key codes used on keyboards normally mapped their actual position not their function. That is why you have to tell the computer the keyboard layout you have. So swapping the actual key codes as well as the graphics was rather odd, IMHO.

ESC swapped with ~ I can just about cope with, but as the key tops swap nicely, I just needed a bit of xmodmap to swap them back...

xmodmap -e 'keycode 9=grave'
xmodmap -e 'keycode 49=Escape'

I agree CAPS LOCK is the most useless key ever, but I cannot see the point of moving it to the left CTRL key position, and I just cannot get used to that. So I want to swap back functionally even if the key tops will not interchange.

Well, using xmodmap as above is fine for these keys as well, except the CAPS LOCK and CONTROL aspects stay where it is! I eventually found that apart from renaming the keys I have to change attributes attached to the keys. First remove them :-
 
xmodmap -e 'remove Lock = Caps_Lock'
xmodmap -e 'remove Control = Control_L'

Then do the remapping...

xmodmap -e 'keycode 37=Caps_Lock'
xmodmap -e 'keycode 66=Control_L'

Then add the attributes back again...

xmodmap -e 'add Lock = Caps_Lock'
xmodmap -e 'add Control = Control_L'

But still - how do I get rid of this strange key that is not a penguin for some reason!

BT listening to us?!

Again actually calling them BT. I know, scary stuff :-)

So what have BT done right?

BT have a DLM (Dynamic Line Management) which tries to fine tune the settings on each broadband line to ensure it remains working at a good speed on a long term basis. The settings can be adjusted in many ways, and they are a trade off. Some settings can make a line more reliable against certain interference but also make the latency higher. Some settings can make the line faster but less reliable. There are a lot of settings.

Now, the DLM, taking care of all of these settings automatically, sounds like a good idea. Unfortunately it has its limits. In some cases the DLM makes matters a lot worse than simply picking some settings and leaving them. It is also impossible to properly tell the DLM what the customer wants - some people want the highest speed, but they know that the line will go off line at high speed every night and don't care as they are asleep! Some people must have a line that is reliable and really do not mind if it is set to only 500K if it works all the time. These are extremes, but especially with reliability the DLM itself causes resyncs and changes in the name of keeping the service reliable! In some cases the line characteristics mean the DLM cycles through different settings over a period of several weeks - thinking it can do better and then finding it cannot over and over again.

BT introduced some changes a while ago that allow the DLM to be slowed down - told not to make changes as often, or even told not to make any more changes. The issue there is that you get stuck with whatever settings you now have.

Now, on BE lines we have no DLM at all. We have direct control of the settings and we let customers tinker as needed. Our customers, or our support staff, can sort lines out if they are a problem and leave the settings where they work well.

So BT have finally listened to us and soon (December I think) we will have the controls to actually make changes to the settings and turn off the DLM on lines. We aim to expose the settings on the control pages so that customers can also make changes.

In most cases the DLM will do fine, and this won't affect most people - but those with iffy lines or special requirements will find this invaluable as a way to get their line working the way they want it.

So, well done BT for listening - I know it has taken a long time but looks like we may finally have something we have been asking for.

Now, can we have cheaper bandwidth please :-)

Friday, 28 October 2011

New blogger not working from ipad

Thank [deity of choice] that I have found the old posting form at last.

The new blog posting form simply does not work from my iPad. Every time, small or large post, it just comes up with a cryptic error.

Hundreds of email addresses in To: line

Yes, it seems out favourite telco have no clue how to send emails. This one person sends emails with hundreds of email addresses in the To: line!!!

I have to scroll down quite a lot to get to the message itself, which is invariably something misusing GMT.

Wanted! Serious amounts of clue.

New serious limitation on all BT broadband lines!

Well, yes, I am saying BT here, because it has been confirmed very very clearly by BT that you should not have more than one PC on a single line. The exact quote is :-

"Openreach engineer has confirmed the same we cannot add multiple PC's to one single circuit as it causes instability."

Initially this appeared as just an engineer note relating to a long 20CN line. We naturally assumed it was just a silly comment, but a team leader confirmed it, initially relating to all 20CN lines, and then apparently checked with Openreach to confirm it and it seems to be any single circuit. It seems it is clear official BT policy that having more than one PC on a single circuit causes instability.

Obviously, from a technical point of view, we cannot see how this could "cause instability" - that simply makes no sense. Well, the whole statement makes no sense. We thought we bought a service that transfers packets (PPP frames) and not something that had strange restrictions on number of PCs. What this means is that if you do have more than one PC connected and an unstable line BT will refuse to fix the fault.

I suspect none of BTs other wholesale or retail customers are aware of this limitation. I would say we'd like a response from BT on this, but given we have asked several times for them to confirm they really mean this, and received the above quote, I am not sure what else these is to say.

I suppose it does mean there is no need for NAT anymore (yay!).

Odd that 10 years ago when BT launched their original broadband services (500K, 1M and 2M) with BT provided routers we were able to order routers with blocks of IPs configured to allow multiple PCs. Strange that BT have introduced this new restriction now.

Thursday, 27 October 2011

Are you sure it is Thursday?

Well, yesterday was a tad strange in that is seems a number of publications that normally happen on a Wednesday did not happen or were later than usual.

But now I am wondering about Thursday - it seems that it is two days, so maybe that is where Wednesday went.

My train ticket clearly states :-
Start date 27-OCT-11
Valid until 27-OCT-11
Validity TWO DAYS

How the hell do they fit two days in 27th to 27th of October?

Wednesday, 26 October 2011

Are you sure it is Wednesday?

www.xkcd.com has not updated
www.cad-comic.com/cad/ has not updated

It can't really be a Wednesday then can it?

Syncing my life to the cloud?

Slightly odd last night at the LONAP social. Being a bunch of geeks almost everyone I spoke to reads my blog. The end result is I had very little to actually talk about with people face to face in the pub.

I imagine people that do facebook a lot have the same problem - when they meet up in real life everyone already knows every tiny detail of everyone else's affairs and there is nothing to actually say.

I was left with discussing the latest blog post I made during the meeting, which people had not yet seen, but generally meant handing over the iPad with the blog post on for people to read. Once done they were synced up with my life and there was not more to say. I did get to see James trying to learn how to solve a Rubik's cube and failing, which was amusing.

What a strange world we live in some times.

[P.S. Blog/rant on Tom's misfortune to follow later]

Tuesday, 25 October 2011

Cannot trust WMBC policers...


This is one for the techies, and mainly for other ISPs, but I'll try and explain it anyway

We buy bandwidth to link to our favourite telco for broadband services. We have to commit to a level of usage in advance for which we pay. We can then use up to that level for that price. We try not to be the bottleneck in our services where possible, but there are events where demand exceeds the level we have. The IOS5 update was an example where we would have needed about 50% extra capacity for a couple of hours. We could allow cases like this to simply use the extra, it the problem is that not only do we pay extra per megabit when that happens, we have to pay for a whole month's worth even if we only use it for 15 minutes. Therefore, like most ISPs, we limit the bandwidth to the commit level, but we increase the level regularly as demand increases generally.

The problem is how to manage the usage so as not to exceed the agreed level. It is different levels for different services on the same link, which makes it complicated. We can do this on the FireBrick. But many ISPs have to rely on our favorite telco to manage the bandwidth for them. They will police the usage for you, and you can choose to police at the commit level, or, say, allow 5% burst or 10% burst or even more if you like. Even though we have our own policers we use them as a safety net setting a 5% burst limit. Many ISPs will set to not allow any burst on the basis that they are charged for a whole month for as little as 15 minutes, so may as well just buy more commit in the first place than make use of bursting.

When this crazy scheme was first announced during the trial we immediately gave them feedback saying they would have a problem if they could not synchronise the policer change with the billing. If they increase the policer one day and the billing later, then they would assume you were incorrectly bursting. They ignored us and we got unexpected burst charges for exactly this, even for over 5% which should not be possible if we have set that as the limit. Bills still in dispute for many months.

So, how do they fix this? Do they start changing the bills before the policer? No, they add a note to the email saying the policer has been changed now, but not to use any extra bandwidth until 11pm as doing so will be charged as burst.

So that means ISPs can no longer rely on our favourite telco to police the service for them. If they up the bandwidth there is a good chance they will get usage to the new limit before the billing change and so pay for a whole month of burst changes.

So this makes the whole point of them policing for you a waste of time. Every ISP has to have their own policers or face expensive burst charges.

All they had to do is ensure billing before policer change for increases (and other way for decreases) and it would be fine.

Arrrrrg!

Monday, 24 October 2011

You can sort our internet out, right?

Always fun when you get known to have some clue on any subject, and sadly when relatives know you have a clue it gets worse... I get a call to come and help sort the Internet for my daughter. She is living in a housing association block of flats with her boyfriend. She suggested that they have no Internet.

So I come along, and get through the security gates (which seem as much about keeping people in as keeping people out) and am shown the common room where, to my surprise, they have a phone socket, netgear wifi router, and extra netgear access point. The router has sync, and the wifi works giving a 192.168 address, but no Internet. Hmmm....

Sadly the staff have no idea at all - the manager is not there - they don't really know who does the Internet, and apparently it has not worked for months. It stops working quite regularly apparently and the residents think that the management don't pay the bill, but who knows? More to the point they have no idea on router passwords or DSL login or anything...

Well, as ever, FireBrick to the rescue, an FB2500 and a Vigor V.120. I put an Apple airport express as a good quality wifi, and set about trying to get some Internet working. They managed to come up with a couple of logins and passwords eventually - but they don't work. Now, thanks to the comprehensive logging on the FireBrick I can see they are on a BT 21CN BRAS (RED6.GI-B). So I try an A&A login, and get a very useful response - our DNS servers, a connection on dummy IP and "Unknown user" message. That means 21CN definitely does not have service selection barring and I can log in to any ISP regardless of who pays for the line - useful. We suspected this for a long time, but nice to see it in practice.

So, setting up on our systems, which usefully tell me the circuit ID on the L2TP connection, and bingo we have fixed IPv4 and IPv6.

Now, when someone works out which ISP they actually use, and works out if they are in a contract, and gets a migration code, we can properly move them over. In the mean time we are providing bandwidth and not a line, and they are paying some other ISP for the line and using no bandwidth. Well, they might be paying, who knows - if the ISP has blocked their login it does not matter now!

In the mean time this is effectively the fastest migration I have ever done, and all with no actual details of the existing service, just physical access to the line. A bit scary really.

Wednesday, 19 October 2011

Is this a good look for me?

Well, the pumpkin head and the light bulb and swirling things - not sure.

What did surprise me is that you can set the WoW currency units for gold, silver and copper.

The command /run COPPER_PER_SILVER=x and similarly for SILVER_PER_GOLD allows the scale of the basic currency display to be changed. It does not change the price of anything, they are displayed based on number of coppers as you have requested, but it means you could run in old style 12 coppers per silver, and 20 silvers per gold if you like!

So now it looks like I have loads of gold (and pumpkin mush for brains) :-)

Monday, 17 October 2011

Nominet and suspending domains

Nominet have said a lot on policy for suspending domains for illegal activity. Essentially, I believe, it all stems from the idea that at present there is an informal arrangement where the police ask for a domain to be pulled and nominet say no, and the police say "you would not be trying to profit from a crime would you?" and the domain is pulled.

OK that is probably a big over simplification, and clearly bullshit as Nominet charge per two years so unless the domain is up for renewal that day they do not profit any more or less by leaving a domain registered at least until renewal.

But the new policies have all sorts of twists and turns and the latest ideas are here.

To me it is simple though.

1. They should normally only take action on a court order - we have a legal process in this country and there is something called due process which means a court should decide, not some private company and not even the police. If there is no court order then nobody, including the police, has any right to interfere in normal legal contractual operations of a company like Nominet or their customers, IMHO.

2. If there is some policy that allows for an expedited process because (Think of the Children) there is some urgent need to take action sooner then fine - I am not 100% convinced, and even if there is a need, then the need if for a means to wake up a judge and get an order sooner, but assuming there is a need then ....

3. An expedited process should have a clear and short time scale - e.g. 24 hours - whatever it actually takes to get a court order, in which time a court order has to be got else (a) the block is reversed, and (b) whoever asked for it has to pay all costs and consequential losses to such an extent that they should have insurance for it in place before they ask.

That is all pretty simple stuff, surely.

It almost does not matter how close the crime is to the domain, or if a sub domain, as long as a court can decide and someone has a chance to defend it. It almost does not matter who is asking for the expedited process (law enforcement or someone else). What it needs is that whoever asks to bypass legal due process is held accountable when due legal process does not agree with them later. Simple!

If not, then action could easily be taken in haste that causes millions of pounds of damage to a corporation, and if that happens someone should pay, and that someone is whoever asked to bypass due legal process. I'd go as far as to say there should be a standard minimum fee for wrongly suspending a domain like £1000 that does not have to be proved as a loss. Anyone bypassing due process risks that and needs insurance for that before they start.

If they are right in blocking domains the risk of costs will not be a problem. If they are concerned over costs they must be unsure of their own processes for ensuring they are right and so should not be allowed to block domains without a court order. They cannot say they are right to ask, and complain at the risk of costs - it is one of the other.

It is not a complicated measure to put in place, and still needs policy on who is allowed to ask, and so on. But it is a simple principle - you get it wrong - you pay!

That is my view anyway.

Flu jabs

OK, being diabetic I end up with a flu jab every year...

We went as a family, well, three of us anyway. Most efficient process ever in the doctor's surgery - in and out in seconds - no waiting. Well done.

It is better for me than not having one, I am sure. Proper flu would be bad, I expect. My rational mind tells me so. And in fact, the needle they use is really thin and does not hurt, well, in my case. My wife did not like the needle and it did hurt.

But really - several days of having a red lump on my arm that is hot, sore, itchy and generally unpleasant - that is not nice.

Ho hum...

Special password rules

Well, there are some people that make some very silly password rules. We have to log in to a system that relates to banking, and they have rules on passwords. What is silly is that in fact a machine has to log in every day to pick up some reports, not a person. As a person I have no reason to even log in, well, except one!

They insist you change your password every few months. This is just crazy, and I have yet to script this, so every three months or so I have to log in and change the password and change the password on the system that does the logging in. I really must automate that!

Now, I would be quite happy if we just used a client cert on the ssl or some such, but no, they use a simple username and password. But why change it? Why not have them allocate a really long complicated password and say that if we use the one they generate then we don't have to change it?

Basically, if anyone did find the password on my systems then, like anyone else forced to change a password periodically, you will almost certainly be able to guess the next password, so forcing a change periodically is pointless. In this instance all I can do with it is download the reports anyway, so not exactly massively high security needed. Heck, why the hell do they not email the reports PGP encrypted, it would be a lot easier!

Anyway, as if that is not special enough - they decided to make a change to the rules for passwords, over a weekend, not tell anyone, and invalidate passwords that don't meet the new format, and not give any clue why you can no longer log in.

It has literally taken me all day to get this sorted, and have my account reinstated, and set a new password.

But, for your amusement, the new rules are :-
  • Must not be the same as your login
  • Must not be the same as any of the past 12 passwords used
  • Must be at least 7 characters
  • Must contain at least 2 numbers which must not be at the start or the end
  • Must not contain 2 consecutive characters the same

It looks like they use upper case letters and numbers only, but they don't restrict to that thankfully, as that would mean they would have more than halved the time needed to brute force a password by adding these rules.

Why on earth do people do this?

I have emailed them http://xkcd.com/936/ but they won't get it!

Saturday, 15 October 2011

Friday, 14 October 2011

iPhone4S Siri (pic)

Correct Horse Battery Staple... It seems Siri won't let you dictate your password, you have to type it!

Thursday, 13 October 2011

Apple melt the internet

Internet traffic last night from around 18:45 to midnight was at unprecedented levels, not only hitting out commit levels for BT 20CN and 21CN but with the unlimited Be connections pushing the total to the gigabit capacity of some of our links. So, not quite melt but none the less a hell of a lot of traffic.

Our upgrade plans are well underway, and we hope to be ready to handle more than a gigabit by Christmas. We did not expect to need anything like that until the Olympics next year so the traffic levels last night were a bit of a shock!

Having been using IOS5 for many months, I can say I quite like it compared to IOS4, with a lot of small improvements and features, especially iMessage.

This is a graph of one of the Internet exchanges we use (LONAP), last night. It is nice to see how well they coped with such a burst in traffic.


Wednesday, 12 October 2011

Think of the children!

Well, it is in then news that the government want blocks on porn to protect children. We are talking legal content being blocked by ISPs, but obviously think of the children and clearly you don't want kids accessing porn? (I do have 5 kids)

Is there a problem to be solved anyway? i.e. does the fact that teenagers have seen porn actually affect them in later life in some detrimental way. I think the problem with answering that is finding any teenager than has not seen porn in order to compare. I am pretty sure younger children are not interested, but you want them not to find stuff by accident as they may find it confusing, distressing, or just icky. Thankfully safe search on the likes of Google already address this for most cases anyway. For older children, they will find a way past any blocking anyway.

The idea seems to be that some of the major ISPs will provide some degree of parental controls, perhaps at the network level. It seems from the ISPs that this is likely to be a free service they offer. Some ISPs already offer such services anyway, so why this is news is unclear, but it seems the big ISPs will be offering this. I say blocking at the network level, but actually the technicalities seem not to have been worked out.

Network level blocking for this is not a good idea technically - for a start, at the network level, there is no way to know who in a household is accessing the Internet. This means blocks will almost certainly be applicable to an entire household. That is going to cause all sorts of problems as the level of blocking that is wanted will be wildly different within a typical family. In practice, the best way to manage any parental controls is on the machines themselves, that is if you think technology is the solution anyway rather than good parenting, or simple policies like young kids only getting to use the Internet in supervised/communal areas.

Not just porn will be blocked, as there is already talk of blocking gambling sites and other adult content. This has all sorts of side effects - apart from the fact that the blocks may apply to an entire household, depending how it is done, the blocks may apply to sites that offer help on gay/lesbian/bi issues, and the like. This is a problem where teenagers do want advice and help without their parents knowing.

What about monitoring? One of the aspects of many existing parental control tools on the PC itself is they not only block, but monitor, allowing parents to see what has been accessed. Well, when you are talking a 5 year old, maybe that is completely acceptable, but what of when it is a 15 year old? or simply other adult members of the household?

Clear lines of responsibility should really be maintained here. An ISP is not responsible for blocking or monitoring the usage of their service - that is something that is well established in many countries and in the EU. ISPs choosing to offer an extra optional service is clearly sensible, but it is worrying that the government think this should be the ISPs responsibility in the first place. Parental controls are the responsibility of the parent not the ISPs, else we really are talking about a nanny state in the real sense.

Blocking is not going to work anyway. One of the key things is that, unless you start working white lists where only specific approved Internet access is allowed (approved list of web sites), no blocking will actually be effective. Parents will just get a false sense of security. Kids will use ssl, vpns and proxies. They do not need to be smart to do this, just type what their mates on facebook tell them to type.

So really, what is the point in all of this actually? It is not clear there is a problem to solve, and it is clear this will not solve the problem. It is all politics for the sake of it, and that is just depressing.

So, well done for large ISPs offering some options. Lets hope they are sensible in what they offer and that it does not have any of the nasty side effects. Personally I would rather see some good PC based tools and good education for parents on what they do, how to install them, and what the side effects are, and actually the ISP simply not involved in this.

Where do A&A stand? Well we have a clear policy - we provide a clean IP connection and we route packets. That is it. We don't monitor or restrict that traffic. When you sign up you are asked to confirm that you want a clean, uncensored, Internet connection. If you don't, then you want another ISP instead. You can, of course, install your own controls on your own machines, or simply talk to your kids and be a good parent rather than try to rely on technology...

Tuesday, 11 October 2011

Popular blog post

Someone please tell me how this works.

Looking at the stats for my blog, the top posts are :-

1. What a moron... This I totally understand - seriously everyone hates junk callers - it has an audio attachment - yes - top post. 120K hits

2. Strange key on my new keyboard? (pic) OK, now I am lost - this was a total silly - a picture of a windows key on my new keyboard. I have managed to escape such keys for many years, but why the hell is that so popular? 70K hits

3. New blogger - has no title on posts. Am I missing ... is even sillier. FFS I could simply not see the "title" part of the new blogger control page. Brain fartingly stupid. So why 28K hits?

If ever I learn how marketing works I'll be dangerous.

Monday, 10 October 2011

IPv6 BMW

There are large companies that still have crappy NAT IPv4 only, but my son's car now has static IPv4 and IPv6, with Ethernet in the centre console and WiFi which even handles AirPlay from the iPhone to his sound system...

Yes, I said he was a tad special at times, and I recall when I was young installing a CB radio on my bicycle handlebars and a 12V battery on the rack. I did not have a BMW :-)

So, how is it done..

1. He has an inverter in the boot - sadly some of the kit works off mains. That does not mean we will not eventually make some DC to DC modules and do it all low voltage but for now an inverter is the simplest way. He has run a control switch to the centre console...

2. He is using an apple airport express as the wifi. Yes, he could use a MiFi for all this but that does not provide either Ethernet or AirPlay. The airport is a good wifi (802.11n) and has a 3.5mm jack to connect to the sound system (which is another story involving amps, subs, jigsaws and MDF). It means he can simply select his car on his iPhone or iPad to play music with no wires involved. Steve Jobs would have been proud.

3. He is using a FireBrick FB2700 and a dongle from t-mobile. Yes, not one of our SIMs as he plans to totally abuse their "unlimited data" and knows I would charge him if he used our SIMs. As expected, the dongle "just works" even with factory default config. That gives him an external (dynamic) IPv4 address - which looks un-NATted even, which is nice, but still only one IP and still only IPv4. He has Ethernet from the FireBrick to allow for someone with an in-car laptop, etc.

4. He is then using a VPN tunnel from the FireBrick to the big FireBrick FB6402 in our office. This allows me to route a /28 IPv4 (not greedy) and a /64 IPv6 (which, of course, includes 1337 as part of the address). No NAT proper Internet connection...

The brick does DHCP on the IPv4 space and Router Announcements on the IPv6.

We have, of course, set reverse DNS on both, and whois data for "James's BMW". I have not gone as far as polling google latitude and setting DNS LOC records, but one day maybe...

He has chosen to leave the WiFi open, but we can lock things down if ever needed on the FireBrick (it is a firewall, after all). Basically, if the dongle he has is unlimited, why not have an open wifi on your car!

So, in short, if you are ever on the motorway (perhaps near Addlestone?) near a black BMW and see wifi KE55ARD BMW (not not actually his number plate, though he would like that) then feel free to use the free in-car IPv6 wifi.

Getting lost

Well, my son can be a tad special at times, and yesterday was a classic example. His navigation skills are sadly lacking, and no amount of technology can help him.

He did in fact have two GPS enabled iPads in the car, and a iPhone, with a passenger that not only could use the iPad mapping stuff, but also, supposedly, knew the way.

However, we just knew to sit there is costa in Harlow and check on latitude and see that he had whizzed round the M25, missing the M4, the A30 and M3 exits, any of which would have worked, and was half way to Addlestone.

Apparently the phone calls were not appreciated, as my daughter (who was following him) had already called and suggested that maybe he should have turned off the motorway by now.

Hopefully he'll grow out of it - I was almost as bad when I was young, cycling to Evesham when I was trying to get to Malvern. But that was in the days before GPS and mobile phones so I have some excuse :-)

Sunday, 9 October 2011

Eating out

Well. I am not normally a restaurant critic, honest, and I know mistakes happen, but this place was such a disaster I felt I had to say something.

The staff did not seem to be interested in complaints...

Over at my sister-in-law's in Harlow, and we decided to go out for a nice Sunday dinner at The Gatekeerer, a Miller & Carter restaurant in Harlow.

Well, for a start they did not actually have a few items on the menu, but only decided to mention this when we started ordering. Sadly it was both items Pauline wanted, which was not a good start.

I had garlic bread which was really not impressive at all, but the BBQ chicken was apparently quite nice.

We then had a huge wait for the main course, in total an hour. The lamb was practically raw. Apparently that is how they "cook" it, but they do not say that when ordering or on the menu or ask how you want it cooked. The steak was really not very nice. Now I like well done steak, which I know is a challenge, but also I know places that can do a nice steak well done. This was charcoal practically. And the pepper sauce was horrid, not sure what they had done to it. The fries tasted like they had been kept warm for the last hour. The jacket potato was not warm enough to melt butter even!

Ok. So not good. We decide to have a dessert and some coffees. They eventually come back to say they have no hot water and cannot do coffee. However the dessert is there ready, do we still want it. 10 minutes later and no dessert.

Ok. Give in, get the bill. Thankfully Sandra checked the bill, as it had a load of things we did not order such as drinks we had already paid for at the bar, and came to £80... So they spend ages sorting it and doing a new bill, and handed us a £96 bill. WTF Eventually a correct £60 bill. No tip left...

I checked the web site and no sign of any way to send them feedback. Maybe this will be suitable feedback.

Thursday, 6 October 2011

An apple a day obviously does not keep the doctor away

RIP Steve Jobs

Lets hope Apple keep going from strength to strength and maintain the same values.

http://www.apple.com/stevejobs/

As for treating cancer with an apple a day, it sounds like that may be almost exactly what he did. Very sad. http://skeptoid.com/blog/2011/10/05/a-lesson-in-treating-illness/

Not forgetting http://xkcd.com/931/ had a good description of how cancer survival works in practice, as well as a special http://xkcd.com/961/ for today...

Wednesday, 5 October 2011

Font sizes

OK, I am expecting to be shot down in flames here - but here goes...

For a start, I fully appreciate that marketing types have very strong ideas on the choice of fonts and sizes and colours and all sorts. If you see a brochure printed or a menu in a restaurant, or anything else marketing have a hand in, then they have chosen the font size, type, style, colour, spacing, everything. They have to - on paper you have no choice and they have control.

However, on a web site, and very specifically a web site that is not marketing but the control pages on a product, that may be different.

My view is that the end user can choose to set the font size they want as a default, which I presume is what they want for the main body of text they see. They pick the font family. They pick background colours even. So having said I prefer 16pt Helvetica, why should a web site overrule my choice - my preference?

We have just had a heated debate that seems to centre around the fact that a lot of (most?) web sites choose to set a font size for the body text that is smaller than default. Some set to a specific font size, which I think everyone agrees is bad, especially anyone disabled, but many set relatively small compared to chosen default font size, e.g. 62.5%.

Why? Is that the right thing to do? Is it right because everyone else does it and so should browsers offer font size choices with perhaps the standard reduction in size as a size in brackets so you can in fact choose correctly?

Should all web sites and all control pages work on body text that is small?

We have had a product for over 10 years that does not set font size and nobody complained in any way - indeed many people said it was a good web interface. Now we have a new product and followed a web designers views (at least so far) for how to do it and the font is smaller. I am trying to clean it up and make the font what the user selected. I am being hounded on irc that I am wrong. Am I?

Should we second guess our users and pick a different size and font family?

Someone has, rightly, pointed out the A&A web site does this. I am now thinking I should fix that. The various control pages we have don't, after all, change font size.

So, what is right here - second guess the users choice, or respect it?

Monday, 3 October 2011

Odd fraud

We have now had yet another letter from the AA thanking us for renewing someone else's cover. This time to a different variation of my wife's name, and for a completely different person.

From what we can tell they are not charging any of our cards or direct debiting any of our accounts, and the AA are saying they are paid.

I can't think what the fraud is that helps anyone saying it was us paying it (well, not quite the right name).

I am obviously not quite devious enough to work out the angle here!

Code libraries

When writing something new there is invariably a key step in the process of do I use a standard library or not?

Now there are a lot of very good long established functions in glibc, and unix/linux generally which I use a lot, obviously. No, what I am talking of here or more esoteric things like processing iCal entries (RFC5545). iCal is just the latest in a long series of examples that almost always come out the same way and mean I end up writing a new library.

One of the problems is that some of the libraries are really bloated when all you want to do is something simple. I had this when I wanted to create a gif or png and the libraries were a nightmare. I ended up making a very simple image generation library which is a doddle to use and very small and fast. The ical library is like that, /usr/include/libical has 45 files totally over 10,000 lines, just for the include files! It took me all morning to find the functions I wanted. Sadly ical time stuff is in fact a pain in the arse to code, so I was really hoping I could use the library. The good thing about a big library like that is that they usually have done a good job.

The problem is that some libraries, especially big ones, can often do nearly what you want. And to use them properly you have to do something extra, or worse, fix the library. Fixing the library gives most of the same problems as having your own library, unless you can convince the author to take on the fix. But the biggest headache is reading and fixing other people's code is never nice - I am sure I am not alone in that!

Well, libical is no different. I thought I could use it, but just trying to use it this morning I have hit some show stoppers. For a start, one of the biggest annoyances, is inconsistency. It seems the parse function does not understand folded line, so I have to unfold the lines first then feed in to the parse - a minor annoyance, but the output function does understand and generate folded lines - why the inconsistency? The next problem is that the parse understands character escaping like \n but the output does not, and so you get syntactically invalid output, and guess what? outlook generate fields with \n in them. So, basically, I have no way to output the ical data in a format that could be parsed by anything. There is more, but that was just enough to convince me I am on to a loser using the library.

What can I say, arrrrrrrg!

Sunday, 2 October 2011

Never knowingly undersold

John Lewis have this long standing tag line "never knowingly undersold" - yet they have on their price commitment that they exclude on-line retailers (and shops not in a high street, or not with a good range of products, or in administration, etc, etc).

That seems to be that they are knowingly undersold, they know this, and they even know where they are undersold (i.e. on-line retailers, etc).

How does anyone get away with blatantly incorrect headline statements in their adverts like this?

It is like "nobody sells at a lower price that us*" (*except those that do). It makes no sense...

I really think that headline statements should not have footnotes that directly contradict the main statement. There is a lot of sensible use of foot notes and small print to clarify and refine statements, but where they directly contradict the main headline statement they should not be allowed, in my opinion.

It is phrases like "never" that at the issue really. Say "never" and you cannot really have small print saying "well, not quite never". Same as "unlimited" and then small print that says there is a limit.

The closest to sensible use of this I saw once was "virtually fat free" something (milk I think, not sure). They were saying that it was very nearly completely fat free but not quite. That is much more honest.

The other thing is that you can make a claim on their price promise, with loads of caveats, but they don't seem to say they will change their ongoing price as a result. That surely has to be the case though as they would, at that point, know they are being undersold. Maybe I should test this and see if they do change their prices in the shop just as a result of being told they are being "undersold".

I am sure when they started the promotion "never knowingly undersold" it meant something. These days it just means (*apart from where we know we are undersold).

Saturday, 1 October 2011

PPPoA DSL users will only want legacy internet

We are still trying to get usable entry level DSL routers with IPv6.

The latest complication here, which really is baffling us, is that Comtrend have decided that they will deliberately disable IPv6 when using the router in PPPoA mode.

The chipset does IPv6, and is the same Broadcom chipset as used by billion. It is not an issue with memory size of anything like that as it will happily do IPv6 if using it in PPPoEoA mode. It looks very much like a deliberate and arbitrary policy decision by Comtrend.

We are trying to get an answer, and all they seem to be saying that there is no demand. That alone makes no sense. There is either no demand for IPv6 in which case why have it in the product at all, or no demand for PPPoA in which case don't have it in the product at all. Saying no demand for some specific combination of unrelated functions makes no sense.

This seems to be totally a marketing thing - and I know my view of marketing departments is a bit DIlbertesque, but this is total madness. As far as I can tell it is not even as if they are going to sell a higher priced product that does do both (which would be an admission that there is demand). So they are taking a no cost choice to disable a combination of unrelated functions for no logical reason just so they can make fewer sales.

Why would PPPoA users only want legacy Internet when PPPoEoA users want IPv6?

Someone please just sell us an entry level DSL router that does IPv6. How hard can it be?!?!?!