Saturday 29th March 2008

I've just finished revitalising the keys of my trusty Logitech Elite keyboard with stickers on the key caps, and as usual the small change in the surface tactility has played havoc with my typing speed and accuracy until my fingers learn the new feel.

As noted elsewhere in Epicycle, something about my typing style or my fingertips wears the characters of modern keyboards surprisingly fast (if not as severely as this talented individual), and in less than a year the most commonly used keys are completely bald. In the past I've bought dead or surplus keyboards of a similar model on eBay and swapped the keys over as required, but back in the autumn a chance discovery while looking for stickers to translate a laptop's US-format keyboard into something more like the UK standard suggested a cheaper and more long-lasting approach.

I've come across a number of keyboard sticker sheets in my time, and most of them have been a disappointment, peeling off in short order and generally lowering the tone of anything they're applied to. This particular brand is different, however, with a very slightly textured matt black background and gloss white letters with a gloss black drop shadow. The overall effect is unobtrusive yet elegant and, based on the set I applied to the keyboard I use at the office back in the autumn, they're impressively long-lived, with no hint of wear or peeling after six months of heavy use hammering out the usual email missives telling management and users why they can't have what they've asked for and why they're stupid or unreasonable even to have asked.

The stickers are a perfect fit on my Logitech keyboard, and although the keys on the Microsoft model I use at the office are less tapered, they look just as good - and, in fact, were a little easier to apply because of the extra wiggle-room provided by the slightly larger top surfaces. I can no longer find the particular eBay vendor I bought them from, which is a pity as I would have no hesitation in recommending him, but this vendor in France appears to be selling the exact same product and at 8 including shipping to England they're still a good buy. For those who share my heavy hands and abrasive fingertips, they're highly recommended.


Thursday 13th March 2008

Last month I related my aborted purchase of a Parrot Rhythm n'Blue car stereo-cum-handsfree unit, and the slightly odd communications I exchanged with AVR Mobiles, but unfortunately since then the experience has taken a distinct turn for the worst. To begin with one of the staff at AVR, who had been very helpful up until then, and had offered to confirm compatibility with alternative handsets after warning so strongly against my Samsung D900, simply stopped responding to my emails. Apparently his colleagues came out in sympathy for a while, too, as I had to send a number of messages about cancelling the order before actually getting a response. Nevertheless, I did finally get a half-hearted acknowledgement of my cancellation (and a promise that the now mute staff member would officially confirm it - this never arrived!) and I hoped that that would be the end of it.

Following a busy ten days or so, however, I realised that I had not received a refund, and after sending a couple more emails I received another promise that it would be investigated. As I write this I still haven't had my money back, but yesterday, to my amazement, the car kit hardware arrived at my desk instead! If I hadn't cancelled the order, this should have been shipped in time for the installation that was booked for February 28th, so sending it out some two weeks later is thoroughly inexplicable. Needless to say I am not pleased about this, as somebody is going to have to pay for it to be sent back to AVR - but as I have yet to receive a response to my latest message informing them of the unexpected delivery it remains to be seen whether they will admit that it is them. I was intending to deal with the company again once the new Parrot RK8200 ships, but at this point the idea of giving them any more of my money is extremely unappealing to say the least...

Meanwhile, partly as a result of AVR's dire warnings about my D900, I started looking at new phones, and unfortunately none of the current offerings come close to filling all of my needs - with even the globally-worshiped iPhone falling short thanks to its lack of 3G connectivity and its on-screen virtual keyboard. I love the Palm OS operating system, in spite of its increasingly visible age, but their phone hardware tends to suck more and more as time goes by and none of the current models have much appeal. In contrast I've never been fond of the Symbian OS, which always seems woefully sluggish and fragile in spite of its obvious flexibility, and as the hardware that uses it is either deeply unsexy, like Nokia's E61, or lacks a proper alphanumeric keyboard like the otherwise very impressive N95, none of those make the grade either. In fact, in my opinion any phone that purports to be designed for email and messaging and yet lacks a keyboard is just plain brain-damaged and I have to confess that I can't see the appeal of these models. What I said about the Symbian OS goes double for the Windows Mobile platform, unfortunately, which rules out the omnipresent HTC handhelds that are usually found re-branded by one's local service provider - which is a pity, as some of them are really rather nice designs. Finally, I've been generally impressed by Karlene's Blackberry 8300, but it fits her needs better than mine and although the latest 8820 models are wonderfully slick with their built-in GPS and Wi-Fi, they still suffer from slow GPRS connectivity and the OS and applications available for it are noticeably more primitive than the matuer and feature rich Palm OS software I'm used to.

So having ruled out everything on the market, my eye somehow fell on Samsung's recently launched SGH-F700. Massively hyped as an "iPhone killer", in fact it bears only a passing resemblance in terms of its general shape and size, and the fact that it uses a touch screen. In reality there is no sensible comparison though - the F700 lacks the impressive fully-featured operating system and browser of the iPhone, and the touch screen is far more conventional and boring than Apple's slick multi-touch interface. Where it scores in my eyes, however, is the presence of a proper slide-out keyboard and a nippy 3.6Mbps HSDPA 3G connection, and given that I'm less interested in taking my entire music collection with me than I am in sending email and browsing the web on the move, those are strong advantages.

I suspected when I bought the phone that it would turn out to be something of a mixed bag, as although the vast majority of the early reviews are overwhelmingly positive, in fact it is abundantly clear that hardly any of the reviewers had actually used the phone in question but were just fleshing out the manufacturer's PR materials. The minor-league reviews are just as bad, with a typical if disappointing example here, intended for real-world user reviews but in fact mostly containing fluff based on other reviews. The contribution of "Niki from Sheffield" is a classic example, as although she awarded the phone five stars out of five, the comment "I look forward to possibly getting my hands on one" clearly indicates her qualifications for doing so... Sheesh!

I've only just had my own F700 back from having the heavily Vodfone-branded user interface replaced with Samsung's considerably more elegant and functional OEM firmware, and as I've barely had time to configure and test the browser and email it's too soon for my own review - but first looks suggest that if this phone was really an iPhone-killer, as so many "reviews" claim, the iPhone must be a lot worse than I'd heard...


12th March

One of the things that has frustrated me during many years of supporting Windows servers is that there is no easy way of changing the boot device driver of an existing installation. In the decade and a half since Windows 95 first implemented the idea, the plug-and-play subsystem has become extremely effective, and once you can boot Windows it will detect and install support for almost any hardware, substituting generic drivers for video cards and IDE adapters if no specific driver is available - but, as the old saying has it, that first step is a doozie...

We've always worked around this by changing the hardware then performing an immediate re-installation of Windows "over the top" of the existing one, which re-runs the PCI bus scan and so updates the boot device driver along with everything else, but although system configuration and application settings are generally preserved this process can have extremely unpleasant side-effects for servers in an enterprise environment. The problems come from Windows subsystems that have been updated since the installation CD being employed, such as the Windows Internal Database that replaced MSDE, later versions of the .Net framework, BITS, and the like, and when these are over-written with the release versions all sorts of unpleasant things happen to installations of WSUS, SMS and other vital parts of our infrastructure.

This has been a particular concern to us recently, as we need to upgrade the disk space in the lightweight PowerEdge 1650 servers that sit in each of our regional offices to handle chores such as print spooling and local distribution of Windows updates, software packages, and antivirus definitions. As part of the project we're installing the 1650's RAID controller daughterboard to replace a software disk mirror within Windows, and it is this that has necessitated the change of boot device driver. When we upgraded the first of the seven servers we simply re-installed Windows as usual, but the subsequent pain suffered by one of my PFYs trying to repair the WSUS downstream server, and by our support company trying to repair the SMS distribution point, was enough to prompt me to search for an alternative technique of repairing an existing installation in-place.

Given that this must be a fairly common procedure when a Windows installation is moved to a new hardware platform, or when a disk controller is upgraded, I was surprised that it hasn't been more widely addressed, but after considerable searching I came across an add-on for the excellent BartPE boot disk system. Written by a US network engineer, the P2V2P plugin scans the PCI bus for disk controllers and, if it recognises one from its internal database, it will copy the correct files into the Windows system folder and then add the appropriate registry entries from the associated .INF file. The add-on requires scripting support within BartPE, which is not included by default, but this can be installed easily enough from other 3rd party plugins here and here.

On the face of it this is exactly what I needed, but as always there is a catch... the BartPE environment seems to have been re-written at some point to remove the requirement for licensed Microsoft components, and as part of this change the Windows Explorer has been replaced with a 3rd party equivalent of the Start Menu. Unfortunately, the VB Script that comprises the core of P2V2P uses an Explorer API to locate the %SystemRoot% folder to allow it to copy driver files and access the registry, and without it the script grinds to a halt with unpleasant error messages.

However, thanks to a long-standing support contract with my favourite independent software house, Synchronous Technologies Ltd, I have a highly-skilled and flexible developer on standby, and as the searchlight that projects the Bat-Signal onto the clouds above Romford is away being cleaned I sent him a quick email instead. On this occasion time was short, so he simply removed the portion of the script that attempted to determine the path of the Windows system folder automatically and replaced it with a hard-coded pointer to the C:\WINNT path in use on these particular systems. This may be a problem in future, as in general only systems that have been upgraded from NT4 or Windows 2000 use that particular path, and the vast majority of our servers are native Server 2003 installations. There are a number of better approaches, and I may well ask SynTech to revisit the script if the opportunity presents itself. One other quirk remains in that the script crashes the first time with an error "File not found in line 117" - this seems to be because it fails to open a file that it hasn't yet copied, and on re-running the CMD file that calls the script it seems to run through correctly the second time. Again, this could doubtless be debugged by a sufficiently practiced eye.

In the event we ran into a few problems with the first live test on Monday, but I traced those to outdated RAID firmware in the PERC controller crashing the disk subsystem when Windows ran the Found New Hardware Wizard after rebooting, and once I had flashed the controller from a boot disk everything worked very nicely indeed. Now that the bugs have been ironed out of the procedure, I'm confident that for each of the remaining servers we can save the hour or so that reinstalling Windows takes, and the subsequent hours of fussing that would be required to repair the damage caused by the reinstallation, and in a department as busy as mine that's not to be sneered at. So my thanks go out to Bart Lagerweij for BartPE, to Jeff Crystal for P2V2P, and especially to Mike Rix at SynTech for rising to yet another challenge at short notice and bolting the various components together into a useable form. You may just have saved what remains of my PFY's sanity.


The EPICYCLE weblog is currently on hiatus, but I expect it will return sometime in 2009. In the meantime, questions, complaints or comments can be directed to me at:



Weblog Archive