Order of the Bath

A couple of weeks ago I ran a long overdue “broken link check” on West Penwith Resources. I do this using the excellent Xenu program which spiders the site and interprets all the HTTP error codes. I know it was over a year since I last did it, but was horrified to discover 25% of the external links broken. There were a few internal bugettes, typos etc. but in the process of fixing the others I came up with these types…

• Web site totally gone, no trace anywhere. I suppose funding may have dried up, companies gone to the wall and, not uncommon among genealogists, owner gone to be with their ancestors.
• Web site moved to another domain. This is understandable if moving up a gear from a freebie ISP host to a “proper job” custom domain name, though a redirect at least on the home page would be a good idea. I moved about 3 years ago, I think, and only recently have I not been able to find links to the old site. My redirect stayed up for most of that time. Big sites change domain when their department name changes or they are taken over. If they must do it, they ought to have the know how and resources to map page to equivalent page on the new site.

  Moving from one ISP site to another makes less sense and makes you wonder if the makers are serious about their site at all. Even if you don’t go the expense of a personal domain name, there are plenty of free hosts available, especially in the genealogy world.

• Individual pages moved and/or changed name. It is not just the amateurs who do this, some big corporate and government sites do it as well. Now listen up site owners—this is a BAD IDEA. Deep linking to internal pages is a matter of fact on the web so think out your structure well in advance and stick with it. We all make mistakes, I have a few pages in odd places, but you need to think very carefully before moving anything.
• A final curse—those sites that die and are taken over by resellers, link farms and other even less desirable content. Good though it is, Xenu can’t always spot these if they don’t use a redirect. The only way to find them is to actually visit your links occasionally. I am fortunate that my site is a working resource so I am using it every day.

I am posting this entry at exactly 10:00 BST = 09:00 UTC, from the Office, where I know the clocks are accurate and reliable (I wrote the piece much earlier). As you will observe from the time of posting above (or below), the Web Host server clocks are far from accurate. Not as far out as I posted on Monday because I misread the time zone charts — PDT is only seven hours behind BST not eight as I thought — but still pretty bad.

Why should this matter? Well, for a start, it looks very silly if the clock on your screen is not right, and you miss appointments. It is also inconvenient. For example you get emails arriving before they are sent and the FTP “upload if newer” features don’t work properly.

But, there are much more important reasons. If you are trying to diagnose a problem then you have to be able to rely on the time stamps in system logs, particularly for network related issues where more than one machine is involved. I am trying to work with the service provider to solve a problem with some emails not getting through. I can send some test mails through at a particular time and they need to look at their logs to see what happened to them. If the time stamps are not right they do not stand a chance of locating my test mail among the many thousand passing through their server. If they ever have to use their logs in a legal case, e.g. prosecuting a cracker, then it will be (should be) thrown out of court if they can’t demonstrate that the clocks were accurate. Perhaps (sly grin) this is why they keep it wrong, so they can’t be forced to produce email evidence for criminal cases?

There is really no excuse for incorrect clocks; there is a perfectly good and free to use system called NTP which can synchronise to any number of publicly available reference systems and, for an internal network, it is straight forward to set up a cascade so that even the humble desktop can be kept within a few milliseconds of mythical “True Time.” For Windows systems, if you should be so afflicted, there are compatible systems available.

Whilst investigating this I came across a related problem which I thought had died out years ago. When setting up WordPress for this blog it asked for my offset to UTC (Coordinated Universal Time en Français) so that the posts would be stamped with my local time (UK). Except that the time it reported as UTC … wasn’t. It was an hour out due to Summer Time (called Daylight Saving Time in some parts of the world). Now I don’t know if WordPress was reporting it wrong, or PHP was returning it incorrectly (unlikely) or the server has been set up badly, but it certainly was not right. When you set the clock on a machine it has to be set to UTC and then any necessary time zone and DST offsets applied. Many systems rely on this being right; for example, EMail time stamps have the UTC (GMT) offset included in them so that at the receiving end they can be converted back to UTC and then to the new local time so it makes sense to the recipient. NTP sorts all this out automatically because it has to work in UTC internally, but if you are relying on the operator;s wrist watch then he is going to set the clock to what he sees and if the time zone has been left at the Zulu factory setting then everything is going to be haywire.

Late Note: I have tested WordPress date reporting and it is OK; PHP function date("H:i T I") returns 10:00 GMT 0 (i.e. no DST) so it is the server that is set up wrongly!

It looks like I am going to have to work at this a bit. Even if I wanted to keep the style as the default, which I won’t, there are clearly some things wrong which need to be addressed.

• There is no author name on the posts. Come to think of it, I didn’t see how to create multiple authors.
• It would be nice to have the time on the posts, even if the server clock is wrong.
• After asking me to specify the format for the date, it has ignored it.
• The Comments link, which shouldn’t be there in the first place, goes to a 404.
• I need a link back to my family page, the West Penwith Resources home, the One Name Study home etc.

Time to RTFM I think.