Remove ugly orphans

If, in your HTML pages, you sometimes get single words at the end of a paragraph and want to make sure that these orphans never exist, here's how to do it.

Orphans can make a paragraph look quite ugly. In effect, all you need is a way of tying together the last word and the penultimate word so that they never separate.

Fortunately, this can be done quite easily. All you need to do is to open up the html page in your favourite text editor. Find the end of each paragraph by opening the search (usually [Ctrl][F]) and looking for '</p>', '</li>' or '</td>' (this will find most instances of end of paragraphs).

In between the last two words, replace the space with '&nbsp;' which is an HTML escape character that represents a 'No-Break SPace'. (HTML escape characters start with an ampersand and finish off with a semicolon.)

The '&nbsp;' character will force the last two words to appear together on the same line.

You can see in the screenshot that in the web page (lower half) that there is enough space for the word 'either' but, because it is 'glued' to the word 'end', it is forced to appear on the next line.

Once you have edited your file, save it as plain text with the same file name.

Snooping IT staff

A recent survey revealed that roughly one in three IT admin staff abuse their privileges by snooping around ordinary user's confidential data such as private files and emails, wages and even your data from the personnel department.

You cannot stop them because they need to have that type of access should you have a genuine problem.

However, you can encourage such abuse to present itself from which point it can be dealt with by the appropriate authorities.

There is a counter-intelligence technique called 'barium meal'. Real Barium Meal (a suspension of Barium Sulphate) is impervious to x-rays and therefore, when present inside an otherwise invisible container - such as your gut - it can reveal the shape of thing such as the container itself or any foreign objects in there. In the image on the right, you can see that it has revealed an obstruction or a narrowing.

In effect, it reveals the presence of a hidden structure.

We need to know the rules so that we can plant something there that will reveal that they have been broken. For example, confidential information must be confidential. So, if we plant something there and it appears somewhere else, we know that somebody has broken the rules (possibly even the Law - Computer Misuse Act, Data Protection Act and so on) and with any luck, the trace back from where it appeared will reveal the perpetrator.

So, we need a way of differentiating between our information and all of the other legitimate traffic.

• Legitimate traffic is caused by reality so one way of differentiating between our traffic and reality is to make it contain false information - preferably a high proportion of false information.
• One problem with false information is is that if it is too far away from what could happen, it will not be believed so, it also has to be plausible.
• Finally, remember that we are using it to identify the person that took it and we cannot do that if it is in a number of places so, it has to be unique.

Essentially, that is it - it has to be unique, plausible but false and you can put it in one place such as a file. If the information ever appears anywhere else, you know that there is a connection between where you put it and where it appeared.

• An example could be a false email where you claim to have lied on your CV about your qualifications although you need to make sure that you can find your certificates first;
• Another could be false information about a senior manager;
• Another could be an orchestrated conversation on the telephone between you and someone else who is in on it. You arrange it so that you both have the same script (passed person to person or some other secure way) and then you play it out over the telephone; and so on.

One thing you need to be, though, is patient.

On the right, you can see that a directory is protected so that only the user has access. In this way, the only other person that has access is someone with root privileges. When they su to root, they can look at and open the files. In that way, if that information ever shows up anywhere you know that it was somebody with root privileges. One thing that you could do with this is to change the data every now and then so that it pin-points a time period.

Wilf Hey and I had some fun with this in the 1990s.

All good fun - for us.

Controlling USB devices

On a Linux system, you can plug your camera into your Linux box and it mounts all right under KDE. However, you might find unmounting it a bit more of a problem and, if you have written to that device, it is best to unmount it rather than just unplug it. So, how do you make it flush its cache before it loses contact?

When you plug in a USB storage device - whether that is a camera or just an SD card or similar block storage device - a KDE daemon opens up a dialogue box that gives you a number of options. The default is 'Open in a New Window'. Pressing [Enter] opens this but removing it safely is a different matter.

So what about the other end of the process - unmounting it?

Now, to remove it properly, just left-click on the icon and then on 'Safely Remove'.

Whereis, Whatis, Apropos

Knowing the name of a program and knowing what it does are always a problem, no matter which operating system you use and whether you use a GUI or a command line.

So, here are some ways out of problems on two UNIX-like systems - Linux (SuSE 10.2) and OpenBSD.

Linux	OpenBSD
What is a program called?
We know that we've installed a browser on the system but we can't remember what it is called. The solution is to use the apropos command, along with a keyword such as, in this case, 'browser'.
I can only remember how the program name starts. How do I find the rest of its name?
If you are using the Bash prompt, all you do is type the first few characters, then press [Tab] twice - it will beep at you the first time.
I've seen another program in that list. How do I find out what it does?
If you know the name of a program and you want to find out what it does without loading the full manual, just use the whatis command.
I need to know where a program that I can use from the bash prompt is.
If you need to know where just the program is, you can use the which command
I need to know where a program and related files such as binary source and manual pages are located
To find the location of these files as well as the actual binary, in Linux, use the whereis command. This command in OpenBSD acts the same as 'which'.
I need to look up details of the program in the manual
The manual shows you a lot of information about a program. Originally, these were the way that a systems administrator refreshed his knowledge of a program. More recently, other ways such as info pages and so on have been used and on some systems the man pages have been neglected to some extent. However, if you are on a text-only terminal, these are a good way of finding out more as they don't need a GUI to work. If you want a GUI, open up Konqueror and type man:program. If you are in a terminal, just use the man command as in 'man program'. Often, you will see a command written as whereis(1) - the number in the brackets is the manual that the manual pages for that command are in (if you don't type that number, the man program will find the first instance of that name (in the lowest numbered manual) and provide you with that. So, you can type 'man 1 whereis' for example.
and, in Konqueror...

Make your own maps

Maps are always nice to have on your website but they are always owned by somebody. Strictly speaking, as far as ownership goes, the map is a work of art and comes under copyright. So, ideally, if you want the copyright of a map, you need to make your own.

Usually, all you need on a website - or even on a piece of business literature - is a fairly basic piece of artwork that shows people the rough, topologically important features so that they will have a fairly good idea of where something is. They don't need something that looks like an aerial photograph. So, let's make our own.

This is how I made one earlier and it shows the positions and views of five photographs taken of the recent (June 2007) high river levels in Derby (UK).

A good source of your area's topography is the photographic images from Google Earth - effectively, this is one of those so-called mash-ups that are so trendy to have at the moment.

First of all, scale and orient the image to suit your needs and then take a screenshot.

Next, open up your favourite, layer-friendly, image editor - we are going to make good use of layers here.

Next, create a transparent raster layer and call it 'Roads'. Select the paintbrush, and draw the relevant roads onto that layer, taking out any unnecessary details - in my example, I had a layer of outline for the roads and filled the roads in on another layer as well.

Next, you need to repeat the new-layer-draw cycle for railways, rivers, key buildings, green areas and anything else that you might need, using different colours.

Note that a tablet is good for this, by the way, and adds more of an artistic feel to it.

Note that you should feel free to stylise the map you are creating as much as you need to - remember that you are not copying the image, you are interpreting it. This is not just lawyer-speak either. At no point in any of this process are you using any automatic tracing tool. For example, if you were copying the image electronically (getting a program to use the information stored in the pixels to produce information in other layers without user interaction), you might be using tools like the magic outline mask tools; colour selection; and so on.

Here, you are doing it all by hand. The only input that the bottom layer - the photograph - is having to the process is that it provides you with a visual guide - it might as well be in a different window as far as the image processor is concerned because it is not making use of it (If you were doing this on a system that used the Beryl or Compiz window managers, you could put Google Earth under your image and make your image transparent - equivalent to tracing paper). Being in a layer in the same image as your map does not contaminate your map with other people's copyright unless you get the image processor to draw your map for you. By keeping it in a separate layer and making sure that it cannot affect other layers directly when you flatten the image, you are making sure that everything that is in the end map is all your own work.

Next, create a white layer ('white ground' in the image on the right) between the screenshot image and your handiwork - this is going to be your background.

Also, add another one above everything you have drawn ('white'). This one is an easy way of toning down your map - all you need to do is just change its opacity. The reason for this is that the purpose of the map is to show people where something is in relation to everything else - a map that doesn't differentiate between various features is of little use here. By toning down the map (which is effectively the background), you are emphasising the bit(s) that you want the user to be interested in.

Next, delete the screenshot layer so now, there cannot be any of the original copyrighted photograph in your image - it is all your own work.

Here, I have added three layers:

• Text - so that each position can be identified. If you were doing this as a map locating your business, you would use your business name;
• Spots - marking the position so that it has a definite place on the map. Again, if this was a business, you would have a block in the shape of your site or the location of the main gate or whatever is appropriate; and,
• Views - here, I have indicated the direction, field of view and roughly the distance of interest covered by the shot.

These are on different layers so that I can change their colours and opacity if I want to, or even tweak their locations by a few pixels if I want. In the screen detail, you can see how they locate (yellow lines) with respect to each other.

Remember, your image does not contain any of the screenshot you took - it is all your own, stylised interpretation. You own the copyright.

Finally, crop the image and save it.

PDF attached emails

You might have been getting a lot of emails recently, that consist only of a PDF file attachment. Of course, you shouldn't open them but just what are they? If it is a virus, why aren't the virus filters filtering them out and if it is spam, why aren't the spam filters filtering them out?

Whilst there are plenty of unexpected file attachments that are clicked on - around 100,000 or so every day that turn people's Windows PCs into zombie servers - the most recent trend is that spammers in pump and dump campaigns are now using PDFs.

Normally, the text in a PDF is written using vectors so that no matter how much you blow it up, you never see pixels as little squares. However, these new PDFs use dark-coloured, square blocks to form letters in the same way that pixels do, and on the same sort of scale.

Blowing up these PDFs reveals them, as you can see. As a result, there are no letters in the file and spam filters fail to filter them out, just little blocks from a specially prepared pixmap.

Even Optical Character Recognition has a difficult task because the letters are all at different heights.

So, we just have to wait for the filter writers to come up with something that works better than the current solutions.

Incidentally, while I have been writing the SuperDVD content (about a week in early July), the spammers have modified their program so that it now sends out attachments that have no file name. One can only presume that this is so that spam filters that look for a small (a few tens of KiB) PDF attachment with some sort of nonsensical name followed by .pdf will now have nothing to go on and therefore they will get through (until the filters are adjusted to quarantine emails that have a no-name file attached - that narrows it down a bit).

Viewing log files in Windows

Some troublesome log files might end up as one line on some text editors and, as it is supposed to be a log file, this could represent a bit of a problem.

The output on the right shows the same log file extract displayed using two Windows text file programs: Notepad; and, WordPad.

Notepad is text-only so you would think that it would be most appropriate to view a text-only file but you would be wrong in this case. Notepad just looks at the file as one large string.

When you drop a file into WordPad, it is able to recognise where the end-of-line characters are and uses them to create new lines for the following pieces of text (although it does go on to make other new lines as well). At least it makes some sense when you look at it which is more than can be said for Notepad.

So, why does this happen?

On the right are screen dumps from Midnight Commander (MC). The top file is from the original log extract and the bottom is the same file that has been copied and then DOSified so that it will read all right in Notepad.

You can see that the DOS version has an extra character (0x0D) and expecting this is the reason why Notepad won't read the UNIX file correctly.

• one is to re-save it using Windows-style end-of-lines; and,
• the other is to try loading it into a spreadsheet such as OpenOffice.org Calc - this is cross-platform so the same applies whether you are running Windows on a PC or Solaris on a Sparc.

To view this using OOo, open the file in Notepad, press [Ctrl][A] to select all and then [Ctrl][C] to copy it to the clipboard (you would do the same in Linux (KDE) by using KWrite).

You can see how the field will be displayed at the bottom of the dialogue box.

With the log file formats, the only space that is included within a field is the space between the GMT/UPC time and the offset (here +0100 for BST). Otherwise, the GET field, referrer, user agent and so on all parse correctly.

Drawing with a tablet

Tablets are coming down in price and people are starting to ask that if they need one, is it really worth spending a lot of money on one (I spied one in a camera shop in Derby for over £200).

On the right, you can see the sort of thing that you can buy from your local supermarket for just a little above £20. This one is a Trust TB-2100 Wireless Tablet and also has a wireless mouse if you want to use that as well.

The sensor size is roughly 4.5"x6" (just a bit bigger than DIN A6 paper size) so a small print will fit under it nicely, as you can see. It also has a plastic sheet under which you can slide the photograph so that the image is protected and held in place.

The pen has two programmable buttons and 2 extra replaceable nibs.

Examples of the type of modification that you can use this signal for include brush colour, opacity, size and hardness.

Colour variation is usually taken as a value from the currently selected gradient and in the image on the right, you can see how size and colour are implemented in The GIMP.

Unless you are using a tablet for mapping large scale images, you will probably find that a cheap one will do everything you want it to.

Hashes in emails

Believe it or not, it is possible to verify that the contents of a file has not changed, without having ever seen it, had it in your possession or had access to it. This can all be done by someone hashing their file and sending you the hash in an email - you can always telephone them to confirm the contents of the hash in order to compensate for email's intrinsic security weaknesses although the original file will not have been compromised.

So, supposing you get a hash in an email, how does this work?

MD5 (Message Digest algorithm 5) is what is known as a trapdoor algorithm. It works by taking a file that you feed it and packing its length so that it is a multiple of 512 bits long. It then takes it, 128 bits at a time, minces it up and then processes that with the next 128 bits until the file is used up. By then, you are left with a 128-bit value that can be represented as 32 hexadecimal values.

It is called a trapdoor algorithm because there is no way that you can derive the contents of the file from which the MD5 function generated the hash. However, if you receive an MD5 hash in email, you can confirm that if a file generates the same hash, the file is extremely likely to have the same contents as the original file.

You can make your own MD5 hashes from a file using the following commands (using 'filename' as the hash source file):

md5 filename
md5sum -b filename
fciv -md5 filename

...for *BSD, Linux and Windows respectively. (Note that you have to download and install fciv from http://support.microsoft.com/kb/841290/ for it to work because this is not part of the default installation.)

You can see in the screenshot that the MD5 algorithm is fast - less than 150ms for a 32MB file. The images in 'bigimage.bmp' and 'bigimage-1.bmp' differ only by 1 pixel which has is only 1 bit different - this giving completely different hashes...

cf5d2bd10954ace727dca6938170c68b and
3e62f783c3207a6c8b1b071ec331fbbb

... and just to show that the file name does not matter, images 'bigimage-1.bmp' and 'bigimage-1a.bmp' are the same and so are their hashes.

MD5 produces a string of 128 bits but there are other algorithms. SHA-1 produces a 160-bit string and together, MD5 and SHA-1 provide the bulk of hashes although it is said that SHA-2 should be the norm by 2010.

More efficient fans

You might have seen cowls attached to the air intakes of air conditioning and cooling units on the flat rooftops of various business units. However this not just some fancy piece of hardware, it does have a serious purpose - it's not to keep the rain off.

The cowls are critical to the efficiency of the fan.

If the fan is using all of the surface of its blades, it will be able to provide a certain pressure drop for a given flow rate. Reducing the extent to which your hardware squanders the pressure drop and the use of the fan's blades is what this is all about.

In addition to the fan efficiency falling because only around 50 per cent of the blade area is in effective use, there is also a higher pressure drop over the intake.

In addition, scaled, fluid dynamic tests have shown that the pressure loss across such an intake is significantly lower therefore the unit as a whole performs substantially better. Using water as a fluid and various shapes of inlet, the coefficient for pressure loss was around 20 times higher for a sharp-cornered intake (as in the no-cowl example shown above-right) than for one with a smooth intake such as the cowl.

As a matter of interested, you can make your air intake consume an even higher proportion of the available pressure drop by having the sharp-cornered design protrude into the space from which air is being taken.

With a reduction in air velocity at the intake, coupled with the virtual elimination of turbulence at the fan, such an assembly is much quieter.

All you have to do is to get the manufacturers to make them.

Remember, you read it here first.

Dye sublimation printing

In the last few years, dye sublimation printing has found its way into small printers at home and also into photo kiosks in supermarkets where the quality of these instant prints is reasonably good.

However, dye sublimation is not limited to printing onto flat, specially prepared paper - it will also print onto irregular surfaces.

Originally, this was done by taking a black and white image and printing it out, laterally reversed, onto a sheet of paper using a special photocopier with the sublimation powder - a hot press transferring it to a tee shirt.

On the right, you can see one that I made in 1991 when I was running a Modern Jazz group.

This was then put through the special photocopier and the process started.

You can see that the definition was all right as well, even though this has been washed literally over a hundred times.

In the heating process, the dye vaporises from the heated paper and this gas then moves to a cooler surface where it turns back into a solid. It doesn't melt at all, it goes straight from solid to gas and back again.

You can see from the letter 'O' that the dye condenses onto the top of the material.

This sweatshirt has lasted years of wear and if you iron with a cool iron, the dye will outlive the shirt.

Whilst these machines are well out of the price range of the average PC Plus reader, you can take an image on a flash card along to a shop and get it printed onto a tee shirt.

You don't have to use images taken with a camera either - you can make up any image (as long as it is legal) and get it printed out.

With your image, aim to get around 150dpi (or thereabouts) with an image size in the order of 12" (30cm) square. Also, remember that the process is aimed at the family snapper so whilst they claim to be able to take any image format, you are limited mainly to JPEG although PNG is worth a try as long as you have a JPEG image as a backup.

If you want to try to find out which image types you can use, incorporate the file type in the name of the file as well as the extension so you might have 'mypicpng.png' and 'mypicjpg.jpg' because, being a Microsoft Machine, if you see the file name at all, you probably won't see its extension.

Have fun.

Computers in films

Computers in films are usually given a visual task that in reality is utter rubbish.

No system designer would send millions of images of fingerprints over a network for a comparison similar to those that we see in police and forensics dramas, for example. In reality, the work is performed on the server which, visually, is far less impressive.

This type of network faux pas, along with many others, can be seen in many films.

However, there is one film that springs to mind that does use a real program.

In Jurassic Park, in a scene where the 10 year-old girl and a few others are trapped by a dinosaur trying to get in, she goes over to a console and utters those famous words; 'This is UNIX! I know this' and starts playing with a 3D file browser.

In the screenshot, you can see the same type of visual output seen in Jurassic park with files as blocks inside directories, all linked together by a tree structure.

• Zoom in and out by holding the mouse middle button whilst moving the mouse up or down; and,
• Rotations are performed by holding [Ctrl] whilst holding the left mouse button and moving the cursor around.

When you click on a node (file or directory), it puts it in the centre of the screen.

If you selected a directory, you can expand it by right-clicking on it and selecting 'Expand' - this opens it up and places it in the tree.

Right-clicking and selecting 'Properties' opens up a dialogue box showing the file/directory's metadata.

FSV has another view, where everything is seen as a progressively-subdivided, 3D block which you can manipulate in the same ways as the tree view.

World map in iris

Here is how to make an image of an eye - or at least the iris part - that has a map of the world wrapped around as the iris. I'm going to use the new Photoshop CS3 on Windows and an old version of the The GIMP (1.2.3) on Linux. Newer versions have more capability but here we shall see how flexible the software is. (Also, my son has commandeered the server with The GIMP 2.2 on it)

1

First of all, you need to acquire a normal image of the earth. The Mercator projection is fairly easy to find on the Internet - you just need to find one without latitude and longitude lines and with the land all of the same colour (as an alternative, you can find one with the sea all of the same colour - it is just so that you can select one or the other).

This is a part of a map that I found (go into Google Images and look for one there - you need one that is maybe five times bigger than your final image because we are going to mess around with it quite a bit.

One thing to bear in mind is that if you are editing for the web (see the film-strip image above-left) then you are only going to be aiming for 70-100dpi whereas, if you are aiming for printed media, you should be aiming for around 300dpi or more.

With this image, you can see that the Greenwich Meridian is not included in the land areas above and below it which makes it ideal (although if you have found one with the water all one colour, that would have been just as good).

However, this has been saved as a JPEG so where there are boundaries - such as the coast line - you have the normal JPEG artefacts.

2

With your image loaded, create two new layers so that we can mess around with masks and colours without destroying the original image before we need to.

With the map layer still having the focus (just click on it in the layers dialogue box), click on 'Select', 'Color Range...' ('Select', 'By Colour...') and in the dialogue, click on the land colour and, if necessary, adjust the fuzziness so that you have all of your land.

Click on 'OK' ('Close') to finish. Now, select the top, empty layer.

3

Select a colour for the land (we'll use green here). In Photoshop and The GIMP, just click on top-left box in the foreground/background tool in the toolbox and the 'Color Picker' ('Colour Selection') dialogue box will open. Select a hue by clicking on the spectrum and select its saturation and value from the large box. Click on 'OK' ('Close') to finish.

4

Next, we want to fill the top layer, inside the mask, with our green.

• In Photoshop, select the flood-fill tool ('Paint Bucket Tool'), make sure that 'contiguous' is not checked and click on the image.
• In The GIMP, just click the mouse on the top-left box in the foreground/background tool in the toolbox and drag the colour onto the image. A useful note is that this flood-fills the mask but it doesn't change the identity of the current tool.

Now, deselect the selected area [Ctrl][D] ([Ctrl][Shift][A])

5

Next, make a suitable blue - this should be darker than the green.

Select the middle layer in the layers dialogue box. Next, fill that layer with the blue in the same way as you did the green.

6

Then flatten it by right-clicking on the image name in the layers dialogue and selecting 'Fatten Image'.

Now, none of the original image remains. (if you were really paranoid, you could delete the bottom layer before you flattened the image.)

7

Next crop the image with the crop tool so that it only contains the part of the map we are interested in (for most of us, that goes from the southern tip of South America up to the top of Russia. Now, we have a longer and thinner map.

8

Next, we want to introduce the area that will become our pupil. This black area needs to take up around 1/3rd of the image. Make black your background colour by clicking on the default colours (the little black and white icon in the colours tool) and then click on the reverse the colours icon to swap them over by clicking on the double headed curved arrow.

Click on 'Image', 'Canvas Size...'.

9p

In Photoshop, click on the bottom-centre of the anchor box and adjust the height so that it is around 25-50 per cent larger.

Then, click on 'OK'.

The map should now appear with black as a solid block at the top.

9g

In The GIMP, type 1.5 into the ratio and drag the image block in the window at the bottom, down to the bottom.

Click on 'OK' and the canvas size will be changed.

However, remember that it is the canvas size we have changed and now, we need to right click on the layer in the layers dialogue box and select 'Layer to Imagesize'.

You will now have the black area at the top.

10p

If you are using Photoshop, you need to make your image square (for some reason, it will make your circular image elliptical if you don't).

Click on 'Image', 'Image Size...', making sure that 'Constrain Proportions' is not checked and adjusting the width so that it is the same as the height.

Click on 'OK' and your image should be square.

and then 'OK'.

Unfortunately, if you want to rotate your image so that something other than the middle of it appears at the bottom, you will have to do this later on (rotate it through an arbitrary angle).

10g

In The GIMP, you don't need to make it square to start with.

Click on 'Filters', 'Distorts', 'Polar Coords...'

Here, you can change the offset angle so that if you want the Americas at the bottom, you can. Select, 'Map Backwards' and 'Map from Top', changing the angle to around 85 or so, depending upon which bit you want at the top.

Note that this will vary according to the Mercator projection map you started with.

11

Now, you need to add an alpha channel (transparency) to your layer. We did start off with a JPEG image - these don't support transparency - so we need to get our image into a transparent layer. There are a number of different ways of doing this.

11p

In Photoshop, use the circular mask ('Elliptical Marquee Tool') to cover the image - this will give us our highlighted circle. Now, we can copy the image and paste it back into the same image which will automatically end up as a new layer, with transparency. Now all we need to do is to right-click on the 'Background' layer in the layers dialogue box, then click on 'Delete Layer' and confirm in the dialogue box that pops up.

11g

In The GIMP, right-click on the layer in the Layers dialogue box and select 'Add Alpha Channel'. Now, select the fuzzy selection tool (the magic wand) and select the two white areas to the side of the iris. Press [Ctrl][X] to cut them out. To get to the same physical state that we are in with Photoshop, just use the crop tool to highlight the entire image, click on 'Auto Shrink' and click on 'Crop'.

Your image should now look something like the one on the right.

12

All we need to do now is start to make it look like an iris/pupil.

In order to do this, we can keep on adding layers that modify the image below until we get something that looks just the way that we want it to. First of all, we might want to adjust the blue and the green to get them roughly how you want them (brightness, saturation) although you don't need to because you can always do that later - it just makes the next steps easier.

So, what layers do we need to add?

Multiple Document Interface (MDI) in Photoshop

Multiple document interfaces (MDIs) have been around for many years - on the PC, they existed in the early 1990s using a 80x50 character DOS interface where they were extremely quick on a processor running at even 16MHz.

On DOS or any GUI, the MDI kept everything in one window and made it easy to control that program's child windows. However, things have changed since then.

Now, we have larger displays and run a number of programs at the same time so, whilst we are editing one image, we could be looking at some instructions or other notes - the nature of the MDI is to take up a whole load of your screen so that you cannot use it for anything else.

Image editors like The GIMP have got around this by using multiple windows so that you can move them around from place to place, mingling with other programs or even sitting on a number of different desktops - something that you don't see on Windows or Mac OS X which are both single-desktop GUIs.

So, how do we change this?

In PhotoShop, make sure that you can see some of your desktop and then, grab hold of an image and drag it out of the MDI. At first, it will slip under the border but keep going and once you've dragged it past the border, it will appear outside the MDI.

For the control windows such as 'Layers', 'Color', the tool box and so on, grab hold of the lighter-coloured area, just inside their borders and drag them - they break away just the same and soon, you can configure it the way you want - complete with other programs positioned so as to make your workspace more functional.

PhotoShop won't let you get away with doing this to all of its windows but once it is the way you like it, click on 'Workspace', 'Save Workspace...' give it a name and click on 'Save'. Next time you open PhotoShop, your windows will either be there or you can click on 'Workspace' and select. The only disadvantage is that when you open a new image, you have to drag it out of the MDI and zoom in - a small price to pay.