As listeners of MacBites will be aware I have been suffering greatly with streaming Flash video problems in the last few months.

The purpose of this post is provide details of the configuration settings I’ve used on a Netgear router to solve my problem, so if you’re affected read on and see if it might work for you.

I first noticed a problem a few months back when I was accessing AdobeTV via the web site or the Adobe Media Player. The streaming video would start to play but after 3 minutes the router reset itself taking up to 20 minutes to reconnect but faster if I manually reset it a second time. This was completely repeatable right down to the exact 3 minutes the stream would play before resetting the router. All the lights would come on the router and these would be static not flashing as usual.

I then noticed that the same thing happened when using the BBC iPlayer – this was now getting beyond a joke. There were anomalies in that using the same physical machine but accessing each site using Windows XP running in Parallels the router wouldn’t crash!

I mentioned it on MacBites a couple of times more as a curiosity than expecting anything to come of it but I couldn’t believe how many people contacted me with similar stories or just to say they have been following my progress with it.

As I said on MacBites episode 14 all is well here now but only because I’ve reconfigured an old Netgear router to connect to Be and stopped using the BeBox. Many people asked about the BeBox with the problem – it’s a SpeedTouch 585 r.7 manufactured by Thomson. Thomson are the manufacturer of BT home hubs so if you’ve got a similar problem but are with BT that might explain it.

There are no more details from BBC/Adobe/Thomson/BT or Be regarding what they’re doing about it if anything so if you’ve got access to another router it might be your only hope at the moment.

My, formerly, spare router is a Netgear GT845PN and I set it up as follows:

1. Backup any existing settings you may need to retain on the router.

2. Reset the router to factory settings.

3. Connect to the router – usually at http://192.168.0.1 using admin/password as the user and password

4. Select basic setup from the menu on the left

5. In answer to the “Does your ISP require a login” Click No

If you have a Static IP Address follow 6.1 if you have a dynamic IP address skip to 6.2

6.1. Select Routed IPoATM and enter your IP Address and Default Gateway as provided by Be

6.2. Select the “Get Dynamically from ISP” option

7. Specify your selected DNS server addresses: Be’s DNS addresses are as follows:

Primary DNS: 87.192.0.51
Secondary DNS: 87.194.0.52

You may elect to use alternative DNS addresses such as OpenDNS (which I use)

8. Click ADSL Settings from the menu on the left, configure the settings as follows:

Multiplexing Method LLC

VPI = 0
VCI = 101
DSL Mode = ADSL2+

And if you’ve got the problem at least you know you’re not alone – I thought I was going mad when I figured out what it seemed to be – assuming nobody else could be having similar problems or that there would be an outcry.

If you’re having problems let me know how you get on using these instructions or just drop me a line to share your pain!!

Disclaimer

First a disclaimer, I am not a fan of spreadsheets. There I’ve said it and I know it’s sacrilege to some.

Don’t get me wrong I use them as much as the next person but I’m always very careful in creating a spreadsheet based solution. As a veteran IT professional I’ve seen too many spreadsheets that should come with a warning slapped on them.

In the best cases they are often over complicated in construction and built without a thought as to what they may need to become in the future. However, in the worst cases they just plain don’t add up. Incorrect references and formula and the most eye straining designs you’ll ever have the misfortune to witness coupled with an obvious misunderstanding of the concept of a spreadsheet all mean too many spreadsheets are just not fit for purpose.

Be it a lack of training or a laissez faire attitude on the part of those who should know better in an organization in my experience it is a pervasive problem in organizations today.
Anyway, climbing down from my soapbox …

History

To add to this dismal situation Excel has a long history of calculation malfunctions!

Just to clarify I don’t mean when a formula is incorrectly formulated by a user. These calculation malfunctions manifest themselves in various situations even when the details entered are completely correct.

What’s Wrong with Excel Now?

Well, the latest issue manifests itself when a collection of positive numbers should add up to a given negative number.

That is easier to show rather than describe so per the following screenshot of Excel, in each of the 3 examples the sum of positive values should add up to the negative value (on Row 1) therefore the result of the SUM in row 10 should be 0 – which as you can see it is not.

For example in column B:

100.03 + 0.04 = 100.07

-100.07 + 100.07 should = 0

But according to Excel it is 7.95891E-15 . If you change the format of the cell to Number and increase the decimal places (see row 12) you can see the REAL values.

So as demonstrated Excel, prior to any forthcoming patches, has a problem with addition and/or subtraction especially evident when calculations involve a mix of negative and positive numbers as the result approaches zero.

Does it matter?

OK so does this really matter? My answer is a resounding “Yes” and in more circumstances than you may imagine.

1. Data Comparison Operations

Any time two numbers are compared you need to be aware that the number displayed on the screen isn’t the complete number Excel uses to make any required comparisons and that complete number might adversely affect the result of the comparison being made.

The most obvious comparison function is the IF statement. For example, an IF function that uses 0 in the first argument, as in =IF(B12=0, “True”,”False”)

However, comparisons in Excel are by no means limited to the IF statement alone. Indeed all the following functions use comparisons:

• IF
• LOOKUP
• HLOOKUP
• VLOOKUP
• SUMIF
• SUMIFS
• COUNTIF
• COUNTIFS
• AND
• OR
• NOT

2. Scaling Errors with Large Multipliers

Another situation where the miscalculation can manifest problems is where extremely large numbers are used. This is because while the errors are many decimal places to the right of a number, multiplications using large numbers will serve to magnify the error.

Why does it happen?

These errors are a result of a floating-point rounding error and if that hasn’t already reduced you to a catatonic state you can read more about the details in Microsoft’s Knowledge Base article 214118.

Could it be avoided?

In a word – Yes and it should be. Programmers often have to write code to take into account failings elsewhere.

A personal allegory will serve as a demonstration.

Many years ago when Excel 2000 was state of the art and Excel 97 was the version of choice for a client of mine I undertook a project to develop an Excel based solution. Part of the solution required a form to be displayed to the user from which they selected a date from an OLE data control.

This worked perfectly on my Excel 2000 system but when deployed at the client’s site the date was constantly 1 day out. The user would select a date using the control but the date displayed was incorrect. After much testing it became clear that Excel 2000 and Excel 97 did not handle dates in the same way when using VBA code to manipulate the data.

Obviously I had to create a solution to work around the VBA anomaly introduced with Excel 2000. I created a simple code block to check the version of Excel in use and adjust the date accordingly.

Wringing my hands and complaining about known anomalies wouldn’t help the client and in this regard Microsoft should take the same approach. While floating-point issues may be the root cause of the problem they can be handled in such a way that the end user is not adversely affected.

A fact that has not escaped other software providers.

Is it only Excel?

Excel is not alone in is it’s miscalculations. Apple’s Numbers calculates incorrectly in exactly the same way.

However, the good news is there are many programs that handle the issue elegantly and calculate correctly, including:

• OpenOffice
• NeoOffice
• Google Spreadsheets
• ZoHo Sheets

In the meantime … what can you do?

There are several techniques available that may eliminate the issue from your spreadsheet.

However, bear in mind that these are only a temporary fix which shouldn’t be needed and may not work in every situation and the burning question how do you know when you’ll need it before the final figures are in?

Sorting

Reordering the listing of values can on occasion solve the problem. Obviously, this just serves to emphasize how ridiculous the problem is! The order in which figures are added should have no impact on the result and there are occasions where reordering items is not possible. For example, a list of sales figures sorted by month can’t be reordered to accommodate Excel’s idiosyncrasies.

Using Precision as Displayed

Another potential solution is the Microsoft recommended use of the ‘Precision as displayed’ option.

‘Precision as displayed’ is an option in the Excel Options the function of which is to force Excel to calculate using only the level of precision displayed in the cell rather than the full data stored within the cell.

This is an approach which should work well with accountancy and financial data where precision to two decimal places is required.

In other situations ‘Precision as displayed’ can actually cause problems of it’s own and certainly the data will be less accurate than with the option disabled.

In Excel 2000/2003 the Precision as displayed option is in the Calculation section of the Excel Options.

In Excel 2007 the Set precision as displayed option is in the Advanced section of the Excel Options.

Using the ROUND Function

The Round() function rounds a value to a specified level of precision. The values contained within the cells are not altered but the value used for comparative purposes is determined by the level of precision specified by the arguments in the ROUND function, as follows:

=IF(ROUND(B12,2)=0, “True”,”False”)

The use of the Round function means that the value contained within cell B12 is rounded down to 2 decimal places prior to it being compared to, in this case, zero thus removing the potential impact of any floating point errors to the far right of the decimal point.

The Round function adjusts a value up or down depending the threshold. If the significant digit is 1,2,3 or 4 it rounds down whereas if the significant digit is 5,6,7,8 or 9 it rounds up.

Conclusion

So there it is – an insight into the latest in a long line of Excel issues to affect the unsuspecting user.

It becomes clear that it is folly to rely on Excel to correctly compute even the most basic calculation. Best advice is check everything … at least twice!

The impetus for this quick tutorial was a post on the MacBites shownotes site from Gazmaz. We’d mentioned our requirements for sharing calendars and Gazmaz wanted to know if there was a way to share calendars on the same Mac but in different user accounts.

After some pondering I gave the following a try and it works perfectly … hope you find it useful.

A couple of weeks ago, on the inaugural episode of MacBites, I mentioned the very useful shortcut feature in Saft and how I’d globalized my shortcuts so they work across all the browsers I have installed. I’ve had so much feedback asking how I implemented this system that I decided to detail the process here.

The shortcut feature was incredibly useful to me when I switched from Opera on Windows to Safari on a Mac as the same feature is built in to Opera and I’d made massive use of it during my time using Opera. The only downside to it was that my shortcuts were marooned in either Opera or Safari and I have a total of 7 browsers installed – Safari, Camino, Opera, Firefox, Netscape, Shiira and Flock (why I have 7 browsers installed is a whole different post!).

So enter Typinator, a utility, from Ergonis Software.

This snippet expanding utility automatically replaces typed text ‘codes’ with longer user-specified ‘snippets of information’. In Typinator a collection of these ‘codes’ is termed a ‘set’

Out of the box Typinator includes an 800 word AutoCorrect dictionary, a real boon for those who constantly make the same typos. However, in addition to this the free TidBITS AutoCorrect Dictionary adds another 2,300 auto-corrections for common mistakes and is available from the Typinator web site.

While both of these ‘sets’ are fantastically useful they, and the Typinator auto-correct features, are not the main subject of this post.

For my global shortcuts we are going to need to create a new blank set of shortcuts, as follows:

Click the add button under the Abbreviation Set list, per Figure 1

Figure 1: Click the add button under the Abbreviation Set list

Specify a name for the new set, URLs in Figure 2

Figure 2: Specify a name for the new set

Now it’s a matter of adding individual abbreviations to expand to full URLs.

To create a new Abbreviation:

Click the add button under the Abbreviation list, per Figure 3

Figure 3: Click the add button under the Abbreviation list

Specify the text you want to type to activate the abbreviation, per Figure 4

Enter the full text in the Expansion area, per Figure 4

Figure 4: Specify the abbreviation and the full text required

There are a few options which add to the configurability of the system, as follows:

Whole Word

There is a check box for specifying whether the abbreviation will expand every time it is typed or only when typed as a whole word.

For example, I might choose to create an abbreviation ‘cook’ which I would like to expand to the word ‘cookery’ if I do not specify whole word this abbreviation will not function correctly. Attempting to type ‘cooking’ ‘cook’ would result in ‘cookerying’ appearing in the text.

Ticking the ‘Whole word’ option corrects this behaviour. Now the word ‘cook’ will only be expanded to ‘cookery’ when it is followed by a space or other punctuation character.

Case Options

There are 3 case options applicable when the abbreviation consists of letters:

1. Case must match – Here for the abbreviation to activate it needs to be typed exactly as it is defined.
2. Case does not matter – Here the abbreviation will activate irrespective of the case used.
3. Case affects expansion – Mixed case in the abbreviation is carried through to the expanded text.

Now the key thing to making my global shortcut system function correctly system wide is that I use a special key as the first character of each abbreviation to ensure that it is extremely unlikely to ever be used in any other application and expand URLs in an inappropriate place. The character I use is the section key (§) just under the escape key on a standard keyboard. So my abbreviation for the BBC news web site becomes §news which in turn expands to the full BBC news URL of http://news.bbc.co.uk.

Another tip is to enable the whole word option for all the abbreviations for your shortcuts. The effect of this is that pressing enter after typing the abbreviation immediately loads in the required web site. The benefits are:

1. Your URL loads in fractionally faster than waiting for the abbreviation to expand and then pressing enter.
2. The number of abbreviations allowed is increased. This is because leaving the whole word option unchecked means that the abbreviation automatically expands and as such each abbreviation must be completely unique with no overlap on the root.

For example, with the whole word option not checked §news would clash with §news2 but with the whole word option enabled on these two abbreviations both would be valid.

All you then need to do is create as many abbreviations as you need for your favourite web sites and you’re good to go!